Saturday 28 January 2017

Belfast, Toulouse, and the Sun

Belfast, Toulouse, and the Sun
Titanic Museum, Belfast

Previous Blog Posts have identified the epidemic of coronary heart disease (CHD) in the UK and also in the USA. Strictly speaking it has been a pandemic, as it occurred simultaneously in all continents, just the tropics having been spared.

The onset of the pandemic in the UK was in the mid-1920s, recorded in a similar way  in all continents. It was described in the UK very clearly and in great detail by Dr Maurice Campbell. His work was published in the British Medical Journal in 1966. At that time the death rate from CHD was still increasing, but it was very close to the peak, and the CHD death rate had just started to fall in the USA. 

The onset of CHD appears to have provoked no curiosity. But in the USA the spontaneous and dramatic fall in the death was very puzzling and people started to think about the reason for it. As a result a Bethesda Conference was convened, to bring together those with the most interest and knowledge. The conference led to the MONICA project (multinational monitoring of trends and determinants in cardiovascular disease).

Although the strict objective was to explain the sudden  reduction of CHD deaths, the identification of its cause was obviously included. The “cause” remained far from clear, and what followed was an industry of “risk factor identification”, reaching a total of 246 by 1981. These gave false messages to the population as most seemed to create a sensationalist headline in the newspapers. This continues today as risk factors tend to be confused with cause, which is elusive.

Variation and susceptibility 

The problem has been the failure to separate “cause” from “susceptibility”. One specific disease will have one specific cause, and this will be international. The incidence of a disease will vary depending on a variety of factors that will influence susceptibility. These can be within communities, families, and nations.

One example is cigarette smoking. It is one of the most clearly defined risk factors for death from CHD. A detailed study of UK doctors showed that 20% of heavy smokers died from CHD, but so did 20% of non-smokers. So where is the risk? The long-term study showed that by 50 years of age, twice as many heavy smokers had died from CHD compared to non-smokers. Also, death from CHD occurs on average ten years earlier in heavy smokers compared to non-smokers. Cigarette smoking cannot be regarded as the cause of CHD, but it is an important risk factor. Furthermore it is a risk factor that can be modified, unlike others such as baldness (to my knowledge the effect of the wearing of wigs by bald men has not been investigated).

Geography

Geography appears to have a major impact on the susceptibility to death from CHD. In the UK there is a gradient of deaths rate from CHD, highest in the north-west and lowest in the south-east. To try to identify an explanation, the Three English Towns Study (which we have seen in a previous Post) failed to do so. There was no significant variation of smoking incidence or dietary pattern. It is generally assumed that the answer lies in differing but unknown behavioural patterns of the regional populations. In other words the victims are assumed to be to blame.


Map of the UK showing mortality rates by region

Location of residence also becomes obvious when looking at the variation of death rates from CHD within Europe: death rates are low in southern Mediterranean countries, and high in northern European countries. The reason is not clear. Greece has the highest prevalence of cigarette smoking but the lowest risk of death from CHD. 

We have seen that the very influential Seven Countries Study pointed at dietary cholesterol as the explanation of variation of CHD deaths. However the conclusion pre-dated the study, which itself was very seriously flawed and is now discounted. Diet does not explain the wide variation.

The explanation of the variation of CHD mortality remains far from clear, but the MONICA study has provided a powerful process of investigation.

MONICA and PRIME 


Belfast: Harland & Wolff shipyard, where the Titanic was built

Rather than investigate large populations in many countries superficially, one off-shoot of the MONICA project was to investigate two smaller populations in great detail. The populations of Belfast, in Northern Ireland, and Toulouse, in the South of France, were chosen. Two papers were published. The first was part of the MONICA project itself, and this studied 366 people in Belfast and 400 in Toulouse. Detailed dietary assessments were based on 80 and 40 people.


The south of France, close to Toulouse

The PRIME study (prospective epidemiological study of myocardial infarction) was established later to investigate the serious problem of excessive premature death of men in Belfast. The lead investigator for PRIME was Alun Evans, professor of epidemiology and public health at Queen’s University, Belfast, Northern Ireland. He and his team undertook a detailed comparison between many population characteristics of the cities of Belfast and Toulouse, in the south of France.

Many epidemiological studies have looked at national statistics, and national death rates in particular. These are readily available but there can be error that is not under the control of the investigator who uses the data. The PRIME study generated its own data, very detailed, collected personally, and with minimal error. The study investigated much larger population samples, 2748 men in Belfast and 2610 in Toulouse.

The Mountains of Mourne, close to Belfast
The first part of the study was to collect health data, in particular death rates from various causes, in the cities of Belfast and Toulouse. Men were studied as the greatest health concern was the apparent early death of men in Belfast. The age groups studied were 45–54 and 55–64, as death within these age groups is considered to be premature. Men dying in these age-groups had contributed to their pensions but did not receive them.


Vineyards close to Toulouse

The second part was to collect data on various risk factors that are generally thought to cause or lead to premature death, such as cigarette smoking, high blood pressure, diabetes, and high cholesterol (it is only in men of these ages that a high blood level of cholesterol gives an increased risk of CHD death, according to the major Framingham study).

The third part was to look at diet, as many people believe that diet is crucial in the cause of CHD. The popular (but seriously incorrect) diet–cholesterol–heart hypothesis is all about diet. The diet study was very detailed and the data was based in individual questionnaires.

Death in Belfast and Toulouse

The differences found were quite remarkable.


Figure 1: Death rates per 100,000 men aged 45–54 years


Figure 2: Death rates per 100,000 men aged 55–64 years

Men in these age-groups living in Belfast were much more likely to die than those living in Toulouse, 40% more likely in 45–54 age group and 85% more likely in the 55–64 age group (all causes of death). This would obviously give a very great concern for those responsible for the health of the population of Belfast.

Looking more specifically, the risk of such men in Belfast dying from CHD was more than four times more likely (>400%) than those in Toulouse. There was a 20% increase in Belfast men dying from cancer.

It was also reported that the myocardial infarction (MI) rates in men aged 35–64 was 781 in Belfast compared to 240 in Toulouse, a three-fold increase in Belfast. 

It is clear that it is much safer to live in Toulouse than in Belfast.

Risk factor analysis

The results of this part of the study are very interesting but are what might be regarded as negative.


Figure 3: Risk factor analysis, men aged 45–54 years

Figure 4: Risk factor analysis, men aged 55–64 years

Firstly, the prevalence of diabetes was higher in the men in Toulouse, nearly three times as great in the older age-group (this could in part be case-finding or definition rather than true incidence). Hypertension and cigarette smoking rates were also higher in Toulouse. The BMI and cholesterol levels were very similar. 

It can be seen in Figures 3 and 4 that the overall prediction of risk of death from CHD was effectively identical for the men in Belfast and Toulouse. This is very different from the reality shown in Figures 1 and 2, indicating the lack of value of risk factor profiles.

There is at present health tendency to predict survival and CHD risk of individuals on the basis of “risk-factor analysis”, undertaken uncritically. It has caused unnecessary concern to several of my friends, and no doubt to countless other people. 

A good friend of mine received a telephone call from her family doctor informing her that she had “a fifty per cent chance of dying”. This came as a shock to her and it certainly created anxiety. However it could be interpreted as good news – that she had a corresponding fifty per cent chance of immortality (assuming that immortality is good news, and this is extremely doubtful). It was the sort of news that came from an over-simplified algorithm rather than from the thought processes of a normal and well-informed human brain.

It is quite clear that classical risk factors do not explain the much higher death rates of the men in Belfast compared to those in Toulouse. It must be emphasised that we are looking at risk of overall deaths, and also specific causes including CHD, stroke and cancer.

Diet

As what we eat is generally thought to be so important in the causation of CHD and other diseases, it became a major part of the MONICA and PRIME investigations. However, in the search for an answer to the high death rates in Belfast, the diet study can only be regarded as a considerable disappointment: diet did not provide the answer.


Figure 5: Dietary characteristics of men in Belfast and Toulouse 

There was a marginally higher total calorie consumption in Belfast, 2340 versus 2295 kcal per day, just 2% higher than Toulouse. There was more protein consumed in Toulouse, less fat, but the major difference was a much more cholesterol in the diet in Toulouse, 50% more than in Belfast.

There was no obvious dietary explanation for the high death rates in Belfast. The population sample in Toulouse consumed more tomatoes, but it would be stretching the imagination for this to be considered of such major health importance.

Conclusion

The purpose of the investigations was to try to explain: “the several-fold differences in risk of coronary heart disease between France and Northern Ireland.”

The result was that: “Neither the classic risk factor scores nor the similarity in major nutrient intake adequately explain the large differences in IHD (ischaemic/coronary heart disease) and other causes of mortality between the centres.” 

And that: “The levels of the classical risk factors found in this study….cannot explain the large differences in the incidence of IHD (CHD) which exist.”

So this exercise was effectively a failure. The studies failed to identify the “cause” of CHD, and they also failed to identify a significant risk indicator. However they clearly identified our ignorance, and this is of vital importance if scientific progress is to be made.

The studies were certainly successful if the purpose was to challenge and then demolish the popular but seriously flawed diet–cholesterol–heart hypothesis. They certainly did that. But, as with so many researchers, the authors stopped short of expressing such a statement. Unfortunately these very important papers, and the important detailed risk factor analysis, were completely ineffective in explaining the causation of CHD and regional death rates. 

In failing to draw attention to these obvious conclusions, the studies were a waste of money. They possible entailed the employment in the two cities of forty people for a period of five years. If the average salary at todays prices would be £50,000 per annum, then the total cost would have been £10 million. The studies have probably been forgotten.

The leaders of the study should have been told by their sponsors to go back and find the answer. In the late 20th century it should not have been acceptable for an 80% excess death rate of men aged 55–64 years in Belfast compared to Toulouse to remain unexplained. We are not living in the dark ages, in which the will of God would have been an acceptable explanation.

How could it be that the search for the answer came to a halt? What a failure of imagination and scientific curiosity. 

Geography

It is clear that living in Belfast is more dangerous than living in Toulouse. Of course Belfast had a period of very serious civil disorder during the years 1969 and 1994. Of a population of 1.5 million, there were 2755 deaths resulting from terrorism, the majority below the age of 40 (a further 691 deaths in people from outside Northern Ireland). Although this was tragic (and in retrospect incomprehensible), the studies described have looked at deaths not from trauma but from CHD and cancer.


The locations of Belfast and Toulouse

The only obvious difference between Belfast and Toulouse is that they are in different places. Belfast lies at 54o 60' north of the equator and Toulouse at 42o 60'. Belfast is thus is about 800 miles (1300 km) north of Toulouse. Both are important industrial cities, Belfast being where the Titanic was built, and Toulouse is the main base of the Airbus construction.

Most simply, geography appears to be of fundamental importance, and the geographical difference is latitude. How does latitude affect the population? The answer is sunlight exposure. 

The further away from the equator is the place of residence, the less is the incident sunlight per year – that is the lower the sunlight intensity or energy at ground level. 

Belfast is at a further disadvantage as it is so close to the North Atlantic Ocean and the depressions that are carried from it by the prevailing winds that follow the Gulf Stream. Belfast has a great deal of cloud cover, and consequently relatively few hours of sunshine as well and low sunlight intensity.

Within the UK we can see that the regions with the lowest death rates (in south-east England) are those that receive the most sunlight. Correspondingly, the West of Scotland has the highest death rates and the least sunshine. 


UK sun exposure map (hours of sun-shine)
It appears to be likely that the higher susceptibility of the population of Belfast to CHD is lack of sun exposure. The main, but not the only, biological effect of the sun on the body is synthesis of Vitamin D. This is known to be important in the development of immunity, and we have seen that immunity is important in protection against CHD. Not only does a reduced level of immunity increase susceptibility to CHD, but increasing immunity is likely to be the reason for the end of the epidemic.

But we can take things a stage further. We have demonstrated in Belfast an increased susceptibility not just to CHD but also to stroke, cancer, and all-cause mortality. It would appear that the sun is of great importance to human health, probably mediated via an increased level of immunity.

Final thoughts

I have communicated with Professor Alun Evans in Belfast concerning this proposal. He was fairly receptive and he told me that at the end of the investigation, those concerned wondered if climate and in particular the sun might have been the critical factor that was being sought but which had not been found. He told me that by then it was too late to include blood levels of vitamin D in the study (blood levels of vitamin D are of course an important index of exposure to the sun).

However the PRIME team did not put this conjecture into print and it has never been followed up as a further and vital investigation. What a wasted opportunity!

There is however a similar but much less comprehensive study of vitamin D and CHD death within France. I will put this into a future Post.

References:

Evans AE, Ruidavets J-B, McCrum EE, et al. Autres pays, autres coeurs! Dietary patterns, risk factors and ischaemic Heart disease in Belfast and Toulouse. QJMed 1995; 88: 469-477.

The PRIME study group (prepared by JWG Yarnell). The PRIME study: classical risk factors do not explain the severalfold differences in risk of coronary Heart disease between France and Northern Ireland. QJMed 1998; 91: 667-676.




Monday 9 January 2017

The lack of association between statin use and CHD mortality changes (BMJ)



The lack of association between statin use and CHD mortality changes.

A paper on this subject appeared in The British Medical Journal (BMJ) Open Edition in 2016:

Time trends in statin utilisation and coronary mortality in Western European countries 
Federico Vancheri, Lars Backlund, Lars-Erik Strender, Brian Godman, Björn Wettermark
(from Italy, Sweden, Scotland) 

http://bmjopen.bmj.com/content/6/3/e010500.fullsid=5eef643f-5336-4632-a42e-6d439b51ecbf
BMJ Open 2016;6:e010500 doi:10.1136/bmjopen-2015-010500


The proposal was to investigate whether a change of statin use during a period of time would be associated with a proportionate decrease in death rate from coronary heart disease (CHD). This is a reasonable proposition in light of the alleged great benefit of statin medications in reducing CHD death rates.
The study compared statin use and CHD mortality between 2000 and 2012, among Western European countries. The countries chosen were Finland, Scotland, Germany, Sweden, Norway, Denmark, the Netherlands, Italy, Belgium, Spain, Portugal and France. 
They all have similar public health systems, with statutory health insurance and universal coverage, based on either direct taxation or income-related contribution. 

The shows a very significant reduction of CHD rates during just 12 years between 2000 and 2012. This is approaching the end of the epidemic. There had already been an 80% reduction in CHD in the UK during the 20 years between the peak in 1970 and 1990.

There was a wide range of CHD mortality reduction, ranging from 26% in Italy to 56% in Denmark, shown in Figure 1. 

Figure 1: Annual CHD death rates, per 100,000 population

The increase in statin utilisation between 2000 and 2012, shown in Figure 2, varied from 121% in Belgium to ten times that, 1264% in Denmark. 

The absolute  statin use in each of the years was calculated as Defined Daily Dose per 1,000 population per day. It is the percentage change that is of the greatest importance. The increase is quite remarkable.



Figure 2: Statin utilisation 2000 and 2012, and percentage change

The specific findings were stated as follows:
  1. there was no statistically significant correlation between statin utilisation and CHD mortality in any of the years between 2000 and 2012; 
  2. there was no significant relation between changes in statin utilisation and changes in CHD mortality over the same period; 
  3. the mean annual change in statin utilisation varied widely across all countries and was unrelated to the annual change in CHD mortality;
  4. countries with similar decreases in CHD mortality rates (France, Italy, Spain and Portugal) had very different increases in statin utilisation; 
  5. similar increases in statin utilisation were observed in countries with very different reductions in CHD mortality rates; 
  6. the countries showed a wide range of statin increase, with no apparent relation to the reduction in CHD mortality; 
  7. there was no evidence that higher statin utilisation was associated with lower CHD mortality;
  8. there was no evidence that a high increase in statin utilisation between 2000 and 2012 was related to a larger reduction in CHD mortality; 
  9. in a Swedish study that included a large sample of municipalities, there was no correlation between statin utilisation and incidence of myocardial infarction or mortality; 
  10. more patients with higher income and educational levels started statin treatment compared to patients with lower income.
Figure 3: Changes in CHD mortality and changes in statin use (multiply by 100 to obtain percentage change)

We can see some important factors from Figure 3.

1 The countries on the left, showing the greatest negative change (reduction) in CHD deaths during the years 2000–2012 are those the had the highest death rates during the epidemic. Those on the right (Portugal, Spain, Italy, France) had the lowest death rates.

2 Portugal, Spain, Italy, France had a similar change in CHD mortality (-3%) but very different increases in statin use (between 2.5 and 8.5%).

3 Finland, Norway, Netherlands and Spain had a very similar change in statin use (about 7%) but very different reductions in CHD mortality (between -3 and -8%).


It was acknowledged that the major decrease in CHD mortality rates in “Western” countries started well before statin therapy became available, an important point that I have made in several posts. The study looked only at data between 2000 and 2012. The peak of CHD deaths was in about 1970 (the onset of the pandemic in about 1924), and statins started on just a small scale in about 1990. The death rate from CHD had dropped by more than half between 1970 and 1990
The paper suggests that this major decrease in death rate was due to “improvement of risk factors, as a result of population-based interventions.”  It also states: “About half of the changes in CHD mortality can be explained by changes in risk factors, whereas the other half are due to treatment or to unknown factors.”  
The specific risk factors underlying this statement are far from clear. Cigarette smoking is clearly one important reversible factor that has been reversed in recent years. But what are the others? Dietary change has not been shown to have a significant effect on CHD mortality. Family history is very important but it is not reversible. Socio-economic status is also of great importance, but it is not easily reversed.
The “unknown factors” are of greatest importance. If we accept a microbiological cause of CHD, then the reason for the fall  in mortality becomes straightforward: it is the development of immunity within the population.
The paper also states that “Clinical trials show that statins lower the absolute risk of coronary death by less than 1% (primary prevention) and 3.5% (secondary prevention). 
Further comments are made concerning the attitude of doctors in respect of the initiation of statin treatment, that it is influenced by factors not directly related to the actual patient risk. The gender and length of clinical experience of doctors is important, also the attitudes of patients to medicines, drug reimbursement policies (in certain countries, unspecified but presumably the USA), prescribing restrictions and marketing by pharmaceutical companies. There is also a comment concerning the influence of the extensive marketing campaigns by the pharmaceutical industries on the decision of doctors concerning statin treatment. 
The paper finishes:
“…. the apparent lack of association we observed between CHD mortality and statin utilisation ….”

“The large growth in statin utilisation over the years does not appear to be related to the coronary risk in the population or to the changes over the years. Factors other than coronary risk, such as demographic changes, socio-economic factors, health-care policies (including educational programmes) and pharmaceutical industry marketing, should be taken into account to explain the rapid increase in statin utilisation among Western European countries.” 

I have pointed out in previous posts that the decline of CHD deaths that marked the end of the epidemic has not been explained by known factors. It occurred before the introduction of statins and any other effective medical intervention. It would appear to be spontaneous, and perhaps because of the development of immunity with the population. 

The value of statins has been greatly exaggerated.