COBERTURA Y USO ACTUAL DE LA TIERRA EN VILLANUVEA

Coronary heart disease can be attributed mostly to classical risk factors and their upstream dietary and lifestyle determinants.

I have discussed in chapters 2 and 3 the evidence base that supports our current understanding of CHD causation. The wide variability observed in coronary heart disease mortality rates cannot be explained mostly by genetic differences, suggesting that environmental exposures during the lifetime are the major causative factors. The atherothrombotic process is well characterized, and the central role of cumulative plaque formation and the development of an active inflammatory and thrombotic environment have been linked to the development of the clinical phase of the disease. The atherosclerotic process has been clearly linked to the cumulative, life course exposure to harmful levels of the major biological risk factors; these are independently associated to coronary heart disease incidence, with clear biological mechanisms and showing consistent effects in different populations. Dietary patterns and specific nutrients are also powerful independent predictors of CHD events, and they might act independently or mediated by downstream biological risk factors. The approaches to control the coronary heart disease burden are based on this risk factor paradigm. This in turn suggests that primordial and primary prevention activities may perhaps represent more powerful strategies that treating patients and reducing case fatality rates. Acute triggers might then play a role by precipitating events in patients with established atherosclerotic lesions.

8.2.1 Rates are not set in stone: dynamics of coronary heart disease mortality rates in England, Scotland, Australia, the Netherlands and Poland

The recent trends for coronary heart disease mortality in younger adults reported in the US, England & Wales, the Netherlands and Scotland UK are disquieting. In England & Wales, the previous falls in age-specific mortality rates appear to be flattening in younger men and women (aged under 55 years). Thus far, rates in older adults continue to decline. In Scotland, a recent period of flattening in young men is also evident. In Australia, the overall decline in age-adjusted CHD mortality rates in Australia since the early 1990s conceals an important change in younger

adults. This recent slowing in the rate of mortality decline is occurring in both men and women aged below 45 years. Interestingly, my analysis of a longer series confirms an earlier observation by Wilson in 1995that hinted at mortality flattening in young Australians.

The adverse trends in some in risk factors rather than deterioration of medical care are the most plausible explanations.

The flattening of the decline in CHD mortality among young Dutch adults in the 1990s is evident. However, the subsequent further decline in mortality rates is reassuring and quite important, mainly because it adds support to the concept that the observed flattening is a real phenomenon and not an artefact of low rates or a trend based on only a few years. The limited data on risk factor trends suggest that they are probably linked to these changes. However, more detailed analyses are urgently needed, and indeed, are currently being undertaken.

The rapid decline in coronary heart disease mortality in Poland after 1990 was a massive natural experiment, with everyone experiencing this decline. The shouldering in Poland in 1990 suggests that this was probably a strong period effect temporally associated with massive changes in diet and lifestyle. Interestingly, there was no evidence of any subsequent slowing down in mortality rates in any age group.

8.2.2 Socioeconomic differences in the rate of decline and shape of the trend: an emerging issue

In the UK, premature coronary heart disease death rates can be three to six times higher in the most deprived groups. They therefore remain a major contributor to social inequalities.

Furthermore, the flattening mortality rates for coronary heart disease among younger adults may represent an early warning sign. The decreasing absolute inequalities and the increase relative inequalities observed in age-adjusted rates suggest that deprivation has a significant role as a trend driver.

In Scotland, the observed flattening in the trends is apparently confined to the most deprived groups. As worse medical management of coronary heart disease in deprived young

adults appears implausible, unfavourable trends in the major risk factors for coronary heart disease, must provide the most likely explanation for these inequalities.

However, the lack of clear socio-economic gradient in the flattening in young adults in England represents an interesting contrast, and merits further consideration. Firstly, the risk factor trends in England have a complex socioeconomic patterning, and the net effect of these diverse risk factor trends might explain the discrepancy with Scotland. Secondly, the CHD modelling in England showed less marked social patterning of the contribution of risk factor changes among young adults. More detailed risk factor trend analyses using a modelling approach by socio-economic status are now urgently needed for Scotland, and for additional countries such as Australia and the Netherlands.

8.2.3 Results potentially explaining the trend drivers

Deaths from coronary heart disease in Poland have decreased rapidly after the great political, social and economic transformation commencing in 1989. This natural experiment offered a unique opportunity to examine the contributions of treatments and risk factors to the observed dramatic decline in mortality.

The major contributors to the Polish mortality fall were large changes in total cholesterol, plus beneficial trends in systolic blood pressure in women and decreased smoking in men. Physical activity also contributed. Together they explained about 55% of the observed mortality decline. Worryingly, adverse trends negated some of these benefits, specifically obesity, diabetes and blood pressure in men and smoking in women.

Evidence-based interventions explained about a third of the mortality fall in Poland. The most important treatment contributions came from therapies for heart failure, angina and secondary prevention. However, it is reasonable to assume that these benefits accrued over the 20 years period, while the abrupt change in diet and lifestyle was occurred within a very short period.

The rapid decline phase of the coronary heart disease epidemic in Poland is thus almost certainly attributable to changes in risk factors that can be linked to the major changes in

dietary factors as a consequence of the socio-political transition to a democracy and a market economy.

A modelling approach can also be used to try and help explained complex trends by socioeconomic status, by crudely quantifying the contribution of risk factors and treatments. In England, the IMPACTsec model suggests that approximately half the CHD mortality fall was

attributable to improved treatment uptake. This benefit occurred evenly across all social groups, which reflects well on the UK National Health Service. Important gains were observed from falls in systolic blood pressure, probably reflecting population level policy interventions on salt. However, the contributions of other cardiovascular risk factors were smaller, reflecting modest recent decreases in powerful risk factors such as smoking and cholesterol. These benefits were then further eroded by continuing and substantial rises in obesity and diabetes prevalence.

8.2.4 Could the rapid changes in CHD mortality simply be artefacts?

These changes in coronary heart disease mortality could be due to artefacts. However, even brief reflection suggests that this appears very implausible. The flattening is consistently seen in similar age groups, has extended over many years or even a decade, and, crucially, has been observed in many countries.

Major changes in coding quality also appear implausible, as most of the countries where the flattening occurred have some of the best death certification systems in the world. The method use to correct rates in Poland was based on the trend observed in cardiovascular disease, usually a better quality level of coding, and similar to other approaches.185 Differential coding precision by age is possible; however this generally tends to occur more frequently in older adults341 , while the flattening has been observed mainly in younger adults.

It can also be postulated that a low rate might show an asymptotic pattern as it approaches a certain “minimum threshold”. Although theoretically possible, any such a threshold must be very low for coronary heart disease based on observations of event rates in individuals with very low overall cardiovascular risk.135,136,342 A threshold based on cases exclusively based on genetically determined events is implausible, as most of the candidate genes and genome-wide association studies suggest that the gene-environment interaction is the main determinants of the expression of harmful phenotypes. 125 Moreover, the Scottish

data shows how rates in the same age-group by socio-economic status (and thus within a similar order of magnitude) might show different trend patterns.

A further key issue is that the trend analysis method used is sensitive to the length of the time periods considered. Therefore, phenomena happening at the end of the time series can be very dependent on the robustness of the rate. However, in the countries where flattening was observed, the period was sustained over several years. Australia and England are interesting cases in this regards, where the flattening was first identified based on a few years’ observations and then persisted when the series were updated.

I have based my observations on the underlying determinants of the trends in Poland and England using a modelling approach. Many assumptions of the model are simplistic, and probably ignore potentially important determinants like life course influences or more complex trends over time. The models also failed to explain all the estimated deaths prevented or postponed. In part this can be attributed to the inherent uncertainty in the model parameters. Reassuringly, sensitivity analysis suggests that the main outputs of the models are not heavily influenced by imprecision. The effect of risk factors is probably underestimated because we assume that their effects are independent. On the other hand, the effect of treatments could be overestimated, particularly because we use effect measures based on randomized clinical trials. Treatment efficacy might be lower in the real world. However, the IMPACT model explicitly takes into account the uptake rates for treatments. Finally, some of the ages specific results are based on smaller numbers and thus, model outputs for age and gender specific analysis are correspondingly less robust.

In conclusion, the rapid changes in CHD mortality trends are therefore very unlikely to be artefacts. Although the role of evidence based treatments cannot be disregarded, the recent changes in mortality trends described in this thesis more plausibly mainly reflect changes in risk factors levels.

However, the conventional understanding of the development of CHD involves a long lag time, spanning decades. The rapid changes observed in recent CHD mortality trends therefore demand a better explanation.