This article and the setting in which the study was carried out reflect very closely my social and political background, my medical training and the way in which I was attempting to fulfil some of my hopes and ambitions. I was born and brought up approximately 15 miles from Cape Town at a time when racial segregation was the norm and when a conscious and unconscious awareness of different racial groups was integral to the society. I went to the University of Cape Town Medical School where wards in Groote Schuur Hospital were racially segregated and where at that time—the mid-1940s—the population of the Cape was White and Coloured (mixed race), with few African migrant workers in the local population. The pattern of disease encountered in the wards was considerably influenced by the population’s racial patterns, with coronary heart disease seen predominantly in the White population. There was considerable interest shown by several researchers in the nutritional and socio-economic background of the disease, John Brock being one of them, popularly known as the ‘Professor of Social Medicine’ not only because of his keen involvement in the dietary and social aspects of disease but also because of his love of tennis! There was also a visit to the Medical School by Ancel Keys, who at that time was beginning to organize his Seven Countries Study, and a great deal of scientific interest in the comparison of disease and the precursors of disease between the different racial groups in the hospital. Brian Bronte-Stewart introduced us to the lipid hypothesis and inter-racial lipid differences, and Clarence Merskey fascinated us with coagulation and lysis differences between the communities.

On graduation at the end of 1951, inspired by The Sick African1 written by Michael Gelfand, then working in Southern Rhodesia (Zimbabwe), I went to Salisbury (Harare) for my internship year in a completely African hospital—and with not a sign of coronary heart disease during my year there. By that time, I had decided to pursue a career working, teaching and researching in and among African populations—in an attempt to achieve what was not possible for me in the racially segregated and oppressive South African setting. I set my goal as working with one of the few ‘colonial’ medical schools at that time: Kampala in Uganda, Ibadan in Nigeria or even possibly Jamaica in the West Indies. Four years postgraduate training in Liverpool and London made me aware of the burden of coronary heart disease in the United Kingdom, and working as Registrar to John McMichael and Jack Shillingford at the Hammersmith Postgraduate Medical School stimulated an interest in cardiovascular disease—but not in the technical aspects of clinical cardiology. Unfortunately, my interest in the nutritional background to atherosclerosis was not shared by McMichael, who regarded any mention of the lipid hypothesis with extreme disfavour. Then, at the end of 1956, I went to Uganda and Makerere College Medical School to join Arthur Williams in the Department of Medicine.

Williams and Jack Davies had shown clinically and pathologically that coronary heart disease was almost non-existent among the African population in Uganda, although Hugh Trowell had reported a single case of coronary heart disease in an African judge. In the Asian community of Uganda, on the other hand, coronary heart disease was extremely common, accounting for almost half of the male deaths in Kampala in 1956–1958. Eager to explore and hopefully validate the nutritional-lipid hypothesis and enthused by the recollection of the work in the Cape Town Medical School, the situation seemed ideal for an examination of serum cholesterol levels and dietary background in African and Asian subjects in the Kampala region. We decided to measure serum cholesterol concentrations in three age levels: 12 years, 20 years and above 40 years. Sadly, this was before the days of a more focused study of fractions of cholesterol and of other lipids.2

The dietary background of most of the African subjects was made up predominantly of staple foods—green plantain and sweet potatoes, cassava, yams and maize with pumpkins, tomatoes and green leafy vegetables. Pulses and groundnuts were used in the sauces, and meat and fish were eaten in small amounts. Daily fat intakes were small, probably 10–20% of total calorie intake. In the Asian community, whatever the religious or dietary group, fats and oils provided a large proportion—probably 30–45%—of the total calorie intake.

Selection of subjects was difficult as detailed in the article. In the African group, 12-year-olds came from boarding schools near Kampala, the 20-year-olds were Makerere students who had recently arrived after several months on their home diet in all three East African countries and the 40-year-olds were drawn from the outpatient ‘minor complaints’ clinics, with a bias towards excluding anyone who looked thin or malnourished. In the Asian group, the 12-year-olds came predominantly from a Muslim non-vegetarian school, the 20-year-olds were students from Kampala schools (two-thirds non-vegetarian) and the adults were selected from patients with minor complaints attending local general practices; those with coronary heart disease or diabetes were excluded (two thirds were vegetarian).

The findings were clear. At all three age levels, the Asian subjects had significantly higher total serum cholesterol levels than African subjects, with levels in the Asian adults (6.4 mmol/l) similar to those seen at that time in American adults. Even at age 12 years, the Asian children had significantly higher levels (5.3 mmol/l) than the African children (4.3 mmol/l). In the adult African subjects, there was no increase in level in the 20-year-olds (4.2 mmol/l) and indeed, in the adults, the level had fallen slightly (3.8 mmol/l), to some degree determined by the inclusion of migrant workers from outside the Kampala region. Vegetarian and non-vegetarian younger Asian subjects (12 and 20 years combined) had identical serum cholesterol levels, whereas in adults, the non-vegetarians had higher levels than the vegetarians (7.1 vs 6.1 mmol/l).

These results were viewed against a considerable amount of data on serum total cholesterol levels in different communities at different ages, and appeared to be in agreement with the conclusion that a rise in serum cholesterol was not a necessary accompaniment of ageing. It also seemed to be in agreement with the general finding that a rise in serum cholesterol with age was associated with a large amount of fat in the diet. The results also suggested that the ingestion of unsaturated and saturated fat in the diet, without regard to the relative quantities, did not prevent a rise in serum total cholesterol.

The concept of an ‘ideal’ level of serum cholesterol was much discussed at that time, a concept that assumed an aetiological relationship between the serum cholesterol concentration and atherosclerosis. A level of ∼4.5 mmol/l was considered by some workers in the early 1950s to be ideal, as somewhat lower levels (∼4.0 mmol/l) found in some ‘primitive’ populations were thought to be conditioned by malnutrition. In conjunction with findings from other parts of the world in the 1950s, we concluded that the blood cholesterol levels in ‘successful’ modern civilizations were biologically abnormal, i.e. detrimental to health, and were apparently an essential precondition for coronary heart disease to become endemic in a population. Although a raised blood cholesterol level did not necessarily cause thrombosis, it undoubtedly predisposed to it.

It is more than 50 years since this article was published, and although it is hard to believe, there is still considerable argument and discussion about desirable levels of serum total cholesterol, and, possibly more importantly, the levels of the various lipid fractions, high-density lipoprotein and low-density lipoprotein cholesterol and the ratios between the various fractions of cholesterol, particularly the total to high-density lipoprotein cholesterol ratio. Although the male population of the United Kingdom still hovers around a mean level of 6.0 mmol/l and recent guidelines recommend desirable levels of total cholesterol of <5.0 mmol/l, it is likely that most practitioners accept levels well >5.0 mmol/l as acceptable. Although a high saturated fat intake and a consequently high serum total cholesterol are probably essential factors for endemic atherosclerosis and coronary heart disease in a population, these may well not be sufficient, and it is evident that many other factors, e.g. smoking, obesity, diabetes mellitus and physical inactivity, can all contribute to risk in susceptible subjects.

This study was carried out in the midst of a programme of clinical work, teaching and laboratory work that made up the day-to-day routine of a medical school appointment, and it was to my surprise that I realized that what I was engaged in was called ‘epidemiology’. I discovered that there was a large body of information about the discipline, and that there was a widespread network of enthusiastic clinical epidemiologists collaborating generously on an international basis. Being in a small medical school in the heart of Africa no longer seemed an isolated position, and with the encouragement of the WHO Cardiovascular Unit led by Zdenek Fefjar, I was soon involved with this new family, all excited about the many prospects for comparative cardiovascular studies around the world.

The studies extended to blood pressure in the many populations in East Africa and, in particular, to several years of close involvement with nomadic groups in Kenya, whose dietary patterns—predominantly milk, meat and blood—made them a unique source of information on the levels of blood lipids and blood pressure. Diabetes mellitus was common, and the absence of atherosclerotic disease in the African diabetic population led to new thoughts on the relationship between the two disorders. Tribal differences in the prevalence of cardiomyopathies added to the pool of problems to be investigated, and there began to be an increasing awareness of the need to learn more of the skills and techniques of the epidemiologist’s trade.

And so I travelled to London to join Jerry Morris in the UK Medical Research Council (MRC) Social Medicine Unit at the London School of Hygiene and Tropical Medicine for the last 5 years of that remarkable group and to learn the trade. This included a 4-year study with Jean Marr of blood lipids in middle-aged London civil servants and our ability to modify their lipid levels by simple dietary advice. Then I went on to the Royal Free Hospital School of Medicine and to the establishment of the British Regional Heart Study, an investigation into the geographic variations in cardiovascular disease in Great Britain, based in 24 British towns. Still on the blood lipid theme, we were able to show that serum total cholesterol levels in all 24 towns were uniformly above the levels then regarded as acceptable/desirable/biologically normal (6.0 mmol/l). Although this may all seem a long way from Kampala and the blood cholesterol levels in African and Asian subjects in three age groups, the pursuit of information that might lead to a reduction in the burden of cardiovascular disease was essentially unchanged.

It would be cheering to think that the 1959 article added to the sum of information that guided, albeit slowly, our thinking over the subsequent half century. Attitudes to diet in the developed countries have changed considerably during that time and particularly in recent years. However, to my personal surprise and disappointment, we still lack a deep commitment to the diet–heart hypothesis, and it is likely that atherosclerosis and its complications will follow us throughout the next half century, almost certainly increasing in those parts of the world that have so far been spared.


Written with warm recollections of my co-author, colleague and friend, Ken Jones.

Conflict of interest: None declared.

* Shaper AG, Jones KW. Serum-cholesterol, diet and coronary heart disease in Africans and Asians in Uganda. Lancet 1959;2:534–37. Reprinted in Int J Epidemiol 2012;41:1221–25.


The Sick African
2nd edn.
Cape Town
Stewart Printing Co.
‘Serum-cholesterol, diet and coronary heart-disease in Africans and Asians in Uganda’
, vol. 
Reprinted in Int J Epidemiol 2012; 41:1221–25