After a slow start, the international community has come to prioritise these conditions, as evidenced by serial High-Level Meetings on NCDs at the UN General Assembly. In 2015, 193 countries committed to reduce premature NCD deaths by one third by 2030, as part of the Sustainable Development Goals. WHO member states have also endorsed a menu of cost-effective NCD best buy policy options that can be used to tackle the pandemic.
that assessed the extent to which 18 NCD policies aligned to the best buys that had been implemented in 151 countries. The reports grouped policies under four time-bound commitments adopted at the second UN High-Level Meeting (appendix pp 2–3). Both reports consist of 151 country profiles and an assessment of whether each of the policies had been implemented fully, partially, or not at all in each country. These assessments were based on national expert opinion, pre-existing data, and policy documents submitted for WHO NCD country capacity surveys. The progress monitors did not provide any global-level analysis of overall policy implementation.
but there has been much less research on why and where effective policies are implemented around the world. Although both WHO progress monitors presented a wealth of information that could be used to explore implementation patterns, there has been no systematic engagement with this rich dataset except for short regional overviews from Europe and the Caribbean.
low human and financial capacity hampering NCD policy implementation in sub-Saharan Africa,
and high levels of social solidarity facilitating the adoption of personally restrictive policies in Scandinavia.
The underlying factors that are commonly cited as affecting policy implementation include region, NCD burden, human and financial resources, and political ideology and social solidarity.
Understanding what makes countries more likely to introduce effective NCD policies is arguably one of the most important issues in contemporary global health research, and it is vital that rigorous quantitative analyses inform global NCD strategy rather than high-profile, n-of-1 examples.
Evidence before this study
WHO produced its first non-communicable disease (NCD) progress monitor in 2015, providing an assessment of the degree to which 151 countries had implemented 18 NCD policies aligned to the best buy interventions. The report did not synthesise national data to produce regional or global summary data. We searched PubMed for any studies that had previously investigated implementation of NCD policies, using the search terms “implement*” AND (“polic*” OR “intervention”) AND (“chronic disease” OR “noncommunicable” OR “non-communicable”). We applied no date or language filters and we hand-searched references to uncover additional studies.
Our search yielded six studies. A mixed-methods Malawian study from 2016 assessed the extent to which the national NCD plan was being implemented in ten health districts, and found that inadequate human and material resources hampered delivery. A 2018 review of NCD prevention policies in five sub-Saharan countries found slow and uneven implementation with notable gaps around physical activity policies. Document reviews and key informant interviews suggested that factors affecting policy adoption included political commitment, human and financial resources, and industry influence. A 2019 analysis of best buy implementation in seven countries in southeast Asia used WHO NCD progress monitor data to assess regional progress. These countries had weak implementation around diet-related policies, but tended to have fully or partially implemented policies for most of the other domains. The authors identified low levels of institutional capacity, funding, intersectoral coordination, and lack of standardised monitoring and evaluation processes as barriers to full implementation. A 2016 synthesis of African WHO country reports found that most countries had partially implemented most NCD policies in 2011, but rates of implementation were falling over time. Policies around diet and physical activity were most widely implemented, while those pertaining to clinical guidelines and cardiovascular therapies were most commonly overlooked. Human and financial resources, as well as high burdens of other diseases were identified as potential barriers to adoption. Two short regional overviews for the Caribbean and WHO European region were also identified from the grey literature. Both provided basic summaries of NCD progress monitor findings for individual countries, and the European document provided basic summary statistics for the region.
Added value of this study
In this geopolitical analysis of 151 countries for which WHO has reported NCD implementation data, we quantified global progress between 2015 and 2017, explored which policies were the most widely implemented, identified countries that were the most and least effective at implementation, and assessed which geopolitical factors were associated with implementation. To our knowledge, this is the first multi-region analysis of NCD policy implementation covering 78% of UN member states, and the first study to quantitively examine national political characteristics and NCD policy implementation rather than health outcomes. On average, countries implemented just under half of the NCD policies recommended by WHO in 2017. Clinical guidelines were the most widely implemented policies and tobacco mass-media campaigns were the least widely implemented. Countries in Europe and central Asia were disproportionately represented among countries with the highest implementation scores, while countries in sub-Saharan Africa were disproportionately represented among countries with the lowest implementation scores. Our multiple linear regression model accounted for around 60% of the variance in policy implementation, using World Bank geographic region, risk of premature NCD mortality, percentage of all deaths caused by NCDs, World Bank income group, human capital index, democracy index, and tax burden as explanatory variables.
Implications of all the available evidence
On average, countries implemented just under half of the NCD policies recommended by WHO in 2017, and implementation is slowly improving over time. Market-related policies, especially those related to alcohol and tobacco mass media, were the least widely implemented, along with the provision of cardiovascular therapeutics. Aggregate implementation scores tended to be highest in high-income countries investing in health care and education. Future research should focus on high-achieving outliers and the nature of the relationships between explanatory factors and policy decisions.
Using the data available in the 2015 and 2017 NCD progress monitors, we aimed to answer the following questions: what is the range and mean number of NCD policies that have been implemented globally? Which policies are the most commonly implemented? Which countries have implemented the highest and lowest number of policies? How has the pattern of policy implementation changed over time? Are there differences in the kinds of policies that are adopted or dropped when comparing countries whose overall implementation scores have risen and fallen over time?
We also aimed to explore the extent to which variance in national NCD policy implementation is explained by commonly cited geopolitical factors: region, NCD burden, human and financial resources, and political ideology and social solidarity.
Table 1Explanatory variables
We used simple descriptive statistics to explore policy implementation across the 151 countries. We then did three sets of analyses. First, we examined patterns of implementation among countries with the top 20 and bottom 20 aggregate implementation scores in 2017 with reference to the rest of the world. We calculated 95% CIs of these mean implementation scores for each policy, using a t distribution for the top and bottom 20 (because of the small sample size) and a normal distribution for the rest of the world.
Second, we examined change in aggregate scores between 2015 and 2017. We divided all countries into three groups based on whether their aggregate score had risen, fallen, or remained unchanged between 2015 and 2017 and produced waterfall charts for each group. A new indicator was added for the 2017 NCD progress monitor: effective mass-media campaigns that educate the public about the harms of smoking or tobacco use and second-hand smoke. As some countries may have implemented tobacco mass-media policies before 2015 we discounted this point when analysing changes in aggregate score between 2015 and 2017 (ie, we set the maximum score at 18 points for both years) to allow for fair comparison.
Third, we assessed the extent to which 2017 aggregate policy implementation scores were associated with region, NCD burden, risk of premature NCD mortality, income group, human capital index, democratic index, and tax burden. We used simple linear regression of the aggregate score on each explanatory variable using SPSS software (version 22.214.171.124; IBM, Armonk, NY, USA). We treated tax burden, income group, and world region as categorical variables with bottom tax quartile, low income, and Europe and central Asia as the references. We also used multiple linear regression to create a model that accounted for all of the variables, and to give an estimate of the effect of a given explanatory variable while keeping the others constant.
Normality was checked with residual plots and found to be satisfactory for all variables (data not shown). In all analyses α=0·05.
Since one third of the countries were missing tax data, we treated tax burden data as an ordinal variable, splitting countries into five bins: four quartiles, and one bin for missing data. Once we had transformed the tax burden data into ordinal values, we found that the remaining missing datapoints were clustered in seven countries. We were unable to find alternative estimates for these countries, and felt that the most robust way to treat the missing data was to remove the seven countries with incomplete data from the regression analyses.
To assess the effect of excluding countries that lacked complete data, we re-ran the multiple linear regression model with these countries included. We also re-ran our analysis of high achievers and low achievers using the top and bottom 30 countries to see if this changed the patterns observed, given that 20 was an arbitrary threshold.
Role of the funding source
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Table 2Top and bottom 20 countries by 2017 aggregate implementation score
Table 3Regression analyses for 2017 aggregate scores and seven explanatory variables
The seven countries excluded from the regression analyses due to missing data were Kiribati, Nepal, Seychelles, Solomon Islands, South Sudan, Tonga, and Vanuatu. We note that these are mainly small island developing states. In the sensitivity analysis we re-ran the model on all 151 countries. All variables remained significant with simple linear regression. The overall model explained 61·1% of the variance in 2017 aggregate scores (p<0·001).
Most countries implemented just under half of WHO-recommended NCD prevention and control policies in 2017. The number of countries that implemented physical activity mass-media campaigns and restrictions on sales and advertising of alcohol fell between 2015 and 2017. There was no change in global implementation of alcohol taxation and mean scores rose for every other NCD policy. Over 70% of countries introduced additional measures between 2015 and 2017. Clinical guidelines, graphic warnings on tobacco packaging, and risk factor surveys were the most commonly implemented policies.
revealed wide variability in the rates with which NCD mortality is declining around the world, and suggested that cardiovascular disease is the main driver of premature mortality in low-income and middle-income countries (although we note that weak mortality data collection systems limit confidence in this finding). Our analysis shows that alcohol measures were very poorly implemented, and while graphic warnings on tobacco packaging are widely used, less than a third of countries have fully implemented tobacco taxation and mass-media policies.
Simple linear regression suggested that high-income democratic countries with high levels of human capital and low rates of premature NCD mortality implemented more policies than low-income, undemocratic countries with low levels of human capital and high rates of premature NCD mortality.
Our original multiple linear regression model explained 60·3% of the variance in the 2017 aggregate policy implementation scores. However, we acknowledge that our regression model is a cross-sectional analysis that identifies associations rather than causality. After removing collinear variables the model explained 55·7% of the variance in 2017 aggregate scores. It is not clear which factors account for the remaining 40–50% of variance in scores and, given the lack of empirical research on this topic we can only speculate. The factors included in our model were poor predictors of change in scores between 2015 and 2017.
This study has a number of limitations. Not all policies are equally effective at combating NCDs and our aggregate scores measure breadth of policies rather than effectiveness. According half a point to cover all degrees of partial implementation is imprecise, potentially rendering radically different policy scenarios as equivalent. We were limited by the available data and followed the approach used by WHO in this area.
We accorded a value of zero to all countries where policy data were not available, reasoning that countries with missing data were more likely to have not implemented the policy than to have implemented them in such a way that WHO could not ascertain their existence. This assumption will underestimate mean scores. Fortunately, missing data only accounted for 3·8% of all policy measures in 2017. Missing data were a bigger problem for the tax burden metric. Our decision to transform continuous tax burden into ordinal categories was a statistical trade-off that allowed us to keep every country in the analysis, at the cost of discarding country-level data.
Many of the indicators are self-reported and data quality varies between countries. Although WHO tries to validate the data, in reality little is known about enforcement. Future research should aim to triangulate country data using multiple sources.
Tax burden is an imperfect proxy for centre-right libertarianism; however, we did not find other measures that quantify this domain for such a large number of countries. The Heritage Foundation philosophically oppose high tax rates, although it is not clear if this bias affects their tax burden figures.
Finally, our geopolitical analysis is an exploratory observational study designed to identify geopolitical correlates, not causes. Our contribution is quantitatively testing common assumptions about geopolitical factors that affect policy implementation.
correlated change in smoking prevalence in 126 countries with implementation of five articles by the WHO Framework Convention on Tobacco Control (FCTC) using a method similar to this study. Hiilamo and Glantz
used regression to assess the association between FCTC ratification and tobacco tax rates for 104 countries, and used the state fragility index
to assess whether implementation was associated with general social and financial development. Juma and colleagues
adopted a qualitative approach to assessing the implementation of NCD prevention policies aligned to WHO best buys, however their work was limited to five African countries. A 2019 analysis by Tuangratananon and colleagues
of best buy implementation in seven countries in southeast Asia used WHO NCD progress monitor data to assess regional progress, alongside other resources. The authors found uneven progress, with implementation gaps largely caused by weak institutional capacity, limited funding, weak intersectoral coordination, and a scarcity of standardised monitoring and evaluation processes. By contrast, our analysis considered macro-level variables on a global rather than regional scale, and carries slightly different policy messages.
and some fiscal policies actually generate revenue.
Nevertheless, high-income countries were over-represented at the top of our implementation table, and crude implementation scores were lowest in low-income countries. Regionally, sub-Saharan countries fared much worse than European countries, even after controlling for social and economic development. Countries that have been lauded for making striking global health gains in recent years, such as Nigeria, Botswana, and Rwanda all came in the bottom ten. This underscores the importance of providing financial and technical support to African countries, especially as they face the highest burdens of premature morbidity and mortality. These findings also highlight avenues for future research, examining the domestic factors that have enabled positive outliers to perform so well. Personal communications suggest that in Moldova, high-level government commitment and intense WHO technical support were important factors in policy implementation.
Recent work by Bollyky and colleagues
found that democratisation explained more variance in NCD mortality than GDP in a sample of 170 countries. To our knowledge, our study is the first to examine national political characteristics and NCD policy implementation rather than health outcomes. We found weak evidence to support the common assumption that democracies outperform autocracies, or that centre-left countries outperform more libertarian countries when it comes to implementing NCD policies. In theory, democracies are more responsive to their populations, however they might also be less likely to impose measures that constrain industry profits, or are construed as limiting personal freedoms.
Looking at countries with reputations for social solidarity versus free-market capitalism, Norway had the joint-second highest aggregate implementation score and Finland came joint fifth, but Denmark, Sweden, Iceland, and the Netherlands did not reach the top 20. The USA performed poorly on market-related policies and came 50th out of 151 overall, however other countries with business-friendly reputations performed well—Singapore came 27th, Ireland 26th, and the UK came joint second alongside Norway.
On average, countries implemented just under half of the NCD policies recommended by WHO in 2017, and implementation is slowly improving over time. Market-related policies, especially those related to alcohol and tobacco mass media, were the least widely implemented, along with the provision of cardiovascular therapeutics. Future research is needed to examine regional and domestic factors that affect implementation, focusing on the over-performing and under-performing countries identified by our analysis.
LNA conceived the study, developed the analytical approach, conducted the analysis, and wrote the first draft. BDN collected data and contributed to the analysis, as well as revising drafts. BYTY helped to conceptualise the initial approach and revised drafts. FG-d-S collected data, contributed to the analysis, and revised drafts.
LNA and FG-d-S work as non-communicable disease consultants for WHO. All other authors declare no competing interests.