2. Literature review

Cutler et al. (Reference Cutler, Deaton and Lleras2006) conducted a comprehensive review of the literature on the causes of mortality reduction in developed countries. The authors identify at least three phases in this process. The first one, between the middle of the 18^th century and the middle of the 19^th century, when a systematic decrease in the mortality rate began to be recorded, which was mainly explained by the increase in economic growth and improved nutrition. In the second phase, from the end of the 19^th century to the first decades of the 20^th century, public health played a leading role. At the beginning, the increase in the concentration of people in cities had negative effects on living conditions and mortality. This fact is known in the literature as “Urban Penalty”. Later, however, improvements in the provision of clean water, waste disposal and public hygiene (washing food, washing hands, boiling milk) had a positive impact on mortality. Finally, the last phase, after 1930, was characterised by developments in medicine: vaccination, antibiotics, medical technology, among others.

While the literature recognises that each of these factors probably played a role in the decline in mortality, there is no consensus on their relative importance in this transition. McKeown (Reference McKeown1976) was a pioneer in arguing that improvement in aqueduct and sewerage facilities, public hygiene and specially nutrition, were the most important factors accounting for the rise of world population since the 18^th century and the decrease in mortality. Samuel Preston (Reference Preston1975, Reference Preston and Easterlin1980, Reference Preston, Caldwell, Findley, Caldwell, Santow, Cosford, Braid and Broers-Freeman1990) argued that improvements in public health and social progress were even more important than economic growth in reducing mortality during 20^th century, especially in developing countries. According to Preston, even in the absence of economic growth, these countries experienced large improvements in mortality and life expectancy indicators. For example, China, which in 2000 had the gross domestic product (GDP) per capita of the United States in the 1880s, had by that year reached the life expectancy of the United States in the 1970s. Several other authors also have highlighted the role of public services, especially the expansion in the provision of clean water and sewerage, as determinants for the reduction in mortality (Cutler and Miller, Reference Cutler and Miller2005; Ferrie and Troesken, Reference Ferrie and Troesken2008; Alsan and Goldin, Reference Alsan and Goldin2015; Önnerfors Reference Önnerfors2015, Jaramillo-Echeverri et al., Reference Jaramillo-Echeverri, Meisel-Roca and Ramírez-Giraldo2019; Gallardo-Albarrán, Reference Gallardo-Albarrán2020).

Fogel (Reference Fogel, Engerman and Gallman1986, Reference Fogel, Rosenzweig and Stark1997, Reference Fogel2004) argues that the decline in mortality in the United States is mainly explained by a higher economic growth that allowed access to better nutrition. As evidence of this, the author shows that, since the mid-18^th century, there is evidence of an increase in agricultural production and the height of adults in that country. People with better nutrition have greater resistance to diseases, particularly infectious diseases, which were one of the main causes of death during the 19^th century and the first decades of the 20^th century. McKeown (Reference McKeown1976), for example, showed that mortality from tuberculosis was reduced by 80 per cent before there was a medical treatment for this disease. However, Soares (Reference Soares2007) argues that gains in life expectancy for both developed and developing countries were largely independent of improvements in income and nutrition.

Cutler et al. explain that, although, in the aggregate, economic growth led to improvements in nutrition, at the microeconomic level, it is not easy to establish the direct link between income and health. Much of the literature supports that better health is what leads to higher incomes. In addition, if there is a causality from income to health, this link goes through education and the differential use of health knowledge and technology. Education, and particularly that of women, is fundamental in the adoption and effectiveness of public health policies, to the extent that they undertake these changes more quickly. Moreover, given that the greatest burden of childcare is on women, mothers that are more educated will also have healthier children.

The role of medicine has become more important in the world since the beginning of the 20^th century, with the advances observed in that field in the developed countries. Prior to that century, there were very few medical treatments for infectious diseases that were widely available. Vaccine research made rapid progress in those years, and, during the first three decades of the century, the vaccines for diphtheria (1923), tuberculosis (1927), tetanus (1927) and yellow fever (1935), among others, were developed. The most outstanding breakthrough of the following decades was antibiotics. In particular, sulphonamides and penicillin were able to reduce mortality at older ages (Mackenbach and Looman, Reference Mackenbach and Looman1988; Cutler and Meara, Reference Cutler and Meara2001; Cutler et al., Reference Cutler, Deaton and Lleras2006; Lesch, Reference Lesch2007; Jayachandran et al., Reference Jayachandran, Lleras-Muney and Smith2010).

In Colombia, empirical evidence on the determinants of the epidemiological transition in the 20^th century is more limited. In this country, until the end of the 19^th century, there was very little social and economic progress. However, the 20^th century was a century of advances in economic growth and in the living conditions of the population (Jaramillo-Echeverri et al., Reference Jaramillo-Echeverri, Meisel-Roca and Ramírez-Giraldo2019). In particular, the first half of the century was characterised by rapid growth in per capita income, while the second half presented greater social investment. Meisel and Vega (Reference Meisel and Vega2007), using a database of more than 8 million observations for people born between 1910 and 1989, show that the average height of Colombians increased by 8 cm for men and 9 cm for women. According to the authors, this increase was mainly the result of the continuous improvement in nutrition throughout the country because of sustained economic growth.

Studies on mortality, epidemiological transition and demographic transition in Colombia have focused on the second half of the 20^th century. Jiménez (Reference Jiménez2014) compiles a database between 1946 and 2001 to make a descriptive analysis of the health transition and its causes in Colombia. The author also discusses the role of health spending and reforms, regional inequalities, the evolution of public service coverage and school enrolment during those decades. Rosselli et al. (Reference Rosselli, Tarazona and Aroca2014) also carry out a descriptive analysis of the decrease in total and maternal mortality in Colombia between 1953 and 2013. The authors highlight the increase in the participation of chronic diseases as the main cause of death and the still high levels of maternal mortality. Diaz and Debón (Reference Diaz and Debón2016) study mortality trends between 1973 and 2005 using data from the Latin American Mortality Database. The authors conclude that the structure of mortality in Colombia in those years was similar to that of developed countries.

Jaramillo-Echeverri et al. (Reference Jaramillo-Echeverri, Meisel-Roca and Ramírez-Giraldo2019) examine the evolution in the living conditions in Colombia over the last 100 years and find significant advances, especially in terms of income, health and education. The authors also use the timing and regional variation of sewerage and aqueduct provision to identify the effect of the expansion of aqueducts and sewerage coverage on mortality reduction in the 20^th century. To this end, they use data from 14 Colombian departments between 1916 and 2014. The results show that achieving 30 per cent aqueduct coverage in the departments reduces gastrointestinal mortality by 5.5 to 7.3 per cent, while sewerage provision reduces the mortality rate by 11 to 12 per cent.

Jaramillo-Echeverri et al. show the positive effect of the provision of aqueduct and sanitary improvements for households in reducing mortality. However, the authors did not examine the characteristics of the epidemiological transition in Colombia nor the behaviour in different age groups.

In this paper, we aim to contribute to this gap in the literature through a descriptive analysis of the epidemiological transition in Colombia by age group and causes of death, as well as a co-integration and structural change analysis to model the long-run relationship between mortality rate and variables denoting nutrition, public health, education, medicine and economic growth.

3. The epidemiological transition in Colombia

In Colombia, as in the rest of the Latin American countries, the epidemiological transition began decades later than in the developed countries. However, it was a much faster process. During the first decades of the 20^th century, there was no change in the fluctuating trend of the mortality rate. It was only until the 1940s that an accelerated decline began. In those 40 years, between 1940 and 1980, total mortality rate fell by 73 per cent. The last years of the century were characterised by low and stable mortality levels. The period of greatest decline was between 1958 and 1978 (see Figure 1 and Table 1).

Figure 1.

Total mortality rate, 1918–1998.

Source: Calculations by the authors based on sources described in the Appendix.

Table 1.

Average annual growth in mortality rate by sub-period

Source: Calculations by the authors based on sources described in the Appendix.

As in the rest of the world, not only was there a generalised decrease in mortality, but there was also a change in the main causes of death. At the beginning of the century, deaths from infectious and gastrointestinal diseases predominated. Pneumonia and diarrhoea accounted for about 20 per cent of deaths in 1938. Other common causes of death were other highly contagious infectious diseases such as malaria, whooping cough, tuberculosis, typhoid fever and measles (see Table 2). The mortality rate for these diseases began to sharply decline between 1938 and 1958 (see Figure 2). For example, mortality from diseases caused by gastrointestinal infections fell 38 per cent in that period; pneumonia, 56 per cent and tuberculosis, 53 per cent (see Figure 2). On the other hand, mortality from cancer and circulatory diseases began to increase.

Table 2.

Percentage of deaths by main causes

Source: Calculations by the authors based on sources described in the Appendix.

Figure 2.

Mortality rate of the main causes of death, 1918–1998.

Source: Calculations by the authors based on sources described in the Appendix.

Over the next 20 years, the transition accelerated much more rapidly. As we have seen before, between 1958 and 1978, the highest rate of annual mortality reduction was observed (see Table 1). Deaths from pneumonia, tuberculosis and diarrhoea fell from 23 per cent of all deaths in 1938 to only 7.2 per cent in 1978. Deaths due to malaria, diarrhoea, whooping cough, typhoid and measles practically disappeared, with a share of less than 2 per cent of the total number of cases. On the other hand, the importance of circulatory diseases, cancer and homicides rose (see Table 2). This latter fact was even more noticeable towards the end of the century, as in 1998, when 52 per cent of all deaths were due to one of these three causes.

The fall in mortality and the increase in life expectancy was followed decades later by the decrease in the birth rate, which changed the age structure of the population in Colombia, especially in the last decades of the 20^th century. This progressive and accelerated ageing of the population led to a larger participation of cancer and circulatory diseases as main causes of death (see Figure 3 and Table 3).

Figure 3.

Population pyramid 1938, 1964 and 2005.

Source: Calculations by the authors based on sources described in the Appendix.

Table 3.

Average annual growth in mortality rate of the main causes of death

Source: Calculations by the authors based on sources described in the Appendix.

In line with it, there was also a transition in the age structure of deaths. In 1918, 39 per cent of all deaths were of people under 14 years of age, while, in 1997, it was 10 per cent (see Table 4). In turn, at the beginning of the century, people over 50 years accounted for only 16 per cent of deaths, but, by the end of the century, they accounted for more than half of deaths (59 per cent) (see Table 4). These changes in mortality by age were not homogeneous throughout the decade, as shown in Figure 4. During the first decades, the only age group that experienced a slight reduction in its mortality rate was young people between 5 and 14 years of age. For the rest of the age groups, the mortality rate even increased. Between 1938 and 1958, mortality decreased for all age groups, but especially for those between 5 and 49 years of age. The period between 1958 and 1978, as mentioned above, saw the greatest reduction in mortality, marked by the decline in infant mortality (under 5 years of age). In the last decades of the century, mortality continued to decline for the youngest, but slightly increased for the elderly.

Table 4.

Percentage of deaths by age group

Note: The missing percentage accounts for non-specified ages.

Source: Calculations by the authors based on sources described in the Appendix.

Figure 4.

Average yearly growth of mortality rate by age group.

Source: Calculations by the authors based on sources described in the Appendix.

From 1948 to 1958, Colombia experience a cycle of political violence as a result of the confrontations between the followers of the Liberal and Conservative parties. After 1958, that level of violence decreased but remained at a higher level that what existed before 1948. However, after the mid-1970s homicides began to increase dramatically as a percentage of total deaths because of drug traffic-related assassinations (see Figure 6).

In this sense, we can identify the stages of the epidemiological transition in Colombia in the following way. The period from 1918 to 1938 corresponds to a pre-transitional scenario characterised by a high and fluctuating total mortality rate. The main causes of death were associated with infectious and gastrointestinal diseases in young people under 14 years of age. Then, three periods of the epidemiological transition in Colombia during the 20^th century can be identified:

• From 1938 to 1958: characterised by the beginning of the decline in total mortality and the mortality rate due to infectious and gastrointestinal diseases. During this period, mortality in the over 5 age groups declined.

• From 1958 to 1978: characterised by the acceleration of the epidemiological transition. Mortality and the share of infectious and gastrointestinal diseases in total deaths drastically decreased. On the contrary, cancer, circulatory diseases and homicides increased as causes of death. In this period, those who benefited most were children under 5 years of age.

• From 1978 to 1998: characterised by the stabilisation of the mortality rate at low levels. In this period, about 38 per cent of total deaths were the result of circulatory diseases or cancer, and 14 per cent from homicides. In fact, in Colombia, during these years, there were almost as many deaths from homicides as deaths from cancer.

As shown in Figure 5, starting in the 1980s, there was a significant difference between the total mortality rate and the non-homicide mortality rate. This difference averaged 5.5 points during the last two decades of the century, and it even exceeded 8 points during the early 1990s. As has been mentioned from 1948 to 1958 Colombia experienced a cycle of political violence. As a result, there was an increase in the participation of homicides in total deaths (3.7 per cent). After 1958, that level of violence decreased (to 2.2 per cent of total deaths) but it remained at a higher level that what existed before 1948 (around 0.7 per cent of total deaths). However, after the mid-1970s homicides began to increase dramatically as a percentage of total deaths (becoming more than 15 per cent) as a result of drug traffic-related assassinations (see Figure 6).

Figure 5.

Mortality rates excluding homicides.

Source: Calculations by the authors based on sources described in the Appendix.

Figure 6.

Homicide deaths as a percentage of total deaths (1938–2000).

Source: Calculations by the authors based on sources described in the Appendix.

4. Social and economic progress in Colombia during the 20^th century

Colombia experienced unprecedented social and economic progress in the 20^th century. The GDP per capita quadrupled between 1918 and the end of the century. It went from being a country with a predominantly rural population to becoming a largely urban country. Living conditions improved with better nutrition, more education and greater access to public services.

During the first decades of the period under discussion, per capita GDP doubled, and the average height of Colombians increased at an annual growth rate of 0.08 per cent, reflecting improvements in nutrition levels. The urban population increased by only 10 percentage points, and there was no significant progress in terms of access to primary education, according to the gross enrolment rate (GER). By 1938, few homes had access to water supply (only 11 per cent).

In the following two decades, the increase in height and GDP continued at a lower rate. The primary education GER almost doubled between 1938 and 1963. Moreover, in the latter year, the urban population already represented 50 per cent of the total population. During these years, the expansion of water supply service also began: from 11 per cent of households to 38 per cent, mainly in urban areas. This social progress was consolidated in the 1960s and 1970s, when there was an expansion in schooling and public services.

As previously noted, the decade of the 1930s and 1940s was particularly important for the progress of medicine in developed countries, with discoveries that quickly moved to developing countries as well. Around 1936, the clinical use of sulphonamide chemical drugs started to grow, and they were effective in reducing mortality from infectious diseases. The development of sulphonamides is known as the pre-antibiotic era, which lasted until the large-scale production and mass application of penicillin in 1941 (Reference Jácome RocaJácome Roca, 2008). Penicillin became the curative treatment for infections such as syphilis, pneumonia, diphtheria, bacterial meningitis and septicemia. According to Jácome Roca (Reference Jácome Roca2008), Colombia quickly entered the penicillin era which first arrived around 1947Footnote ¹.

Finally, although the last two decades of the 20^th century in Colombia represented the period of greatest violence in the country, even during these years, there was evidence of positive economic growth and consolidation of social progress (Figure 7).

Figure 7.

Evolution of variables related to social progress in 20^th-century Colombia.

Source: Calculations by the authors based on sources described in the Appendix.

5. Empirical strategy

The second part of this paper focuses on identifying the factors associated with the decline in total mortality in the period from 1918 to 1998. For this purpose, econometric techniques for the analysis of non-stationary time series are used. Specifically, a co-integration analysis is performed, and a vector error correction (VEC) model is estimated. Finally, a structural change test is conducted, in order to measure the impact of the use of sulphonamide drugs and penicillin on the reduction of pneumonia and tuberculosis mortality.

For this analysis, a database with information on mortality by cause and by age group, schooling, economic growth, water supply coverage and height, for the 1918–1998 period, is compiled. Height is used as a proxy for nutrition levels. For this purpose, population and death and birth records from the 20^th-century censuses in Colombia (1938, 1951, 1964, 1973 and 1993 censuses) are used. We also use information from statistical yearbooks since 1918, which consolidate important information on births and deaths by age and cause of death. The series of GDP per capita is estimated by GRECO (study group of Colombian economic growth by its acronym in Spanish) (2002) of the Central Bank. And finally, the height series corresponds to the one consolidated by Meisel-Roca et al. (Reference Meisel-Roca, Ramírez-Giraldo and Santos-Cárdenas2019) from court certificates.