Fax: (011) 334 7273 8650
Increasing mortality rates of common malignancies in Colombia
An emerging problem
Article first published online: 20 SEP 2004
Copyright © 2004 American Cancer Society
Volume 101, Issue 10, pages 2285–2292, 15 November 2004
How to Cite
Piñeros, M., Hernández, G. and Bray, F. (2004), Increasing mortality rates of common malignancies in Colombia. Cancer, 101: 2285–2292. doi: 10.1002/cncr.20607
- Issue published online: 29 OCT 2004
- Article first published online: 20 SEP 2004
- Manuscript Accepted: 19 JUL 2004
- Manuscript Revised: 7 JUL 2004
- Manuscript Received: 11 MAY 2004
- Instituto Colombiano para el Desarrollo de la Ciencia y Tecnologia-Colciencias
- Terry Fox Run Award
- past mortality trends;
- prevention and control;
- data quality
As a result of a major social and demographic transition, Colombia is currently undergoing major changes in disease-specific mortality rates, including an increasing burden of cancer death. The current article described some aspects of the evolution of cancer mortality in Colombia and, in particular, highlighted the trends for the most common causes of cancer death in Colombia.
Cancer deaths registered in the national mortality database from 1981–1996 were used to obtain age-standardized mortality rates by gender and site using the world standard population. The estimated annual percentage change was obtained by fitting a simple log-linear model to the rates in the last decade of recorded data, to gauge recent and near-future cancer mortality trends.
Between 1987 and 1996, the most common causes of cancer death were gastric carcinoma (17% of all cancer deaths), followed by lung (10.5%), prostate (6.2%), cervical (6%), and colorectal carcinoma (5.4%). There were observed declines in mortality trends noted in both genders for gastric carcinoma. Trends in lung carcinoma appeared to be reaching a plateau among men, but increased among women. There were apparent increases in rates of death from colorectal carcinoma for both genders, and from prostate carcinoma for men. Cervical carcinoma appeared to be increasing moderately.
The authors reported that mortality rates from the most common malignancies are increasing, indicating that effective strategies for cancer control need to be put into immediate practice. To monitor and evaluate future trends, the provision of high-quality data also needs to be addressed. Cancer 2004; © 2004 American Cancer Society.
As in other Latin American countries, Colombia is undergoing a major social and demographic transition. Major changes in overall mortality rates in recent decades include a continuing decrease in infant mortality rates for both genders, and an increase in mortality rates among young men age < 45 years. These factors, in combination with population ageing and population growth, have meant that an increasing proportion of the Colombian population is elderly (age ≥ 65 years), and for the greater part, female. Such changes in the population age structure will have important effects on disease patterns and demands on health services, and consequently on health expenditure.
The burden of disease, as measured by person-years of life lost, indicates that for persons age ≥ 45 years, 75% of years of life lost are due to chronic diseases, and that the loss is higher among persons living in urban rather than in rural settings.1, 2 As in many countries, there are increasing trends towards urbanization, with slightly > 70% of the population now living in urban areas.3
Although violent deaths continued to rank high among Colombian men in the last decade, cancer mortality rates has been increasing over this time. Although there are important differences in mortality rates according to geographic region, overall deaths in Colombia due to malignant tumors have moved from eighth place in 1964 to third place in 1996.4
Colombia has limited data regarding cancer incidence because only the cancer registry covering the urban area of Cali city currently provides a long-term data series. In the current article, we describe the evolution of cancer mortality for Colombian men and women, highlighting, on the basis of recent trends in the most common causes of cancer death, possible explanations for the observations.
MATERIALS AND METHODS
Cancer mortality data and population denominators were obtained from the national vital statistics office at the Departamento Administrativo Nacional de Estadística (DANE) for the period 1981–1996. The trend analysis was limited to that period to maintain a consistent classification system throughout the series, and to focus the investigation on more recent mortality trends. In addition, the mortality data only became available electronically after 1979. Causes of death for the whole series have been classified by DANE using the International Classification of Diseases, 9th revision.5
The proportion of deaths where information on age is missing in the DANE database was high overall (1.1%) but relatively stable over time; therefore it seemed acceptable to eliminate these deaths from the analysis. Unspecified deaths from uterine diseases (ICD-9 179) were “corrected” by reassigning them to deaths due to carcinomas of the cervix uteri (ICD-9 180) or corpus uteri (ICD-9 182), according to the proportion of cervix and corpus deaths by 1-year and 5-year age group.
To permit international comparisons, age-standardized rates (ASR) have been computed using the world population age structure of Doll after Segi.6 As an indication of the magnitude of change in the trend, the estimated annual percentage change (EAPC) of the adjusted rate was estimated by fitting a simple regression model to the log of the ASR in the last 10 years (1987–1996) using Stata Version 7.0 software (Stata Corporation, College Station, TX).7
Overall Mortality Rates
There was an average of 19,400 cancer deaths annually between 1981 and 1996, of which only slightly more occurred among women (52%) than among men (48%). Also, 5.6% (17,332) of the cancer deaths in the series had been certified by nonmedical practitioners, a considerably higher proportion than that observed in other countries.
Table 1 shows the average annual number of cancer deaths by gender for the most common malignancies in two time periods, 1981–1983 and 1994–1996. The most common causes of cancer death in both genders combined during the last 3-year period were gastric carcinoma (17% of all cancer deaths), followed by lung (10.5%), prostate (6.2%), cervical(6%), and colorectal carcinomas (5.4%). Although gastric carcinoma deaths have decreased and prostate carcinoma deaths have increased, the general ranking of cancer sites did not change substantially over time.
|153/154||Colon and rectum||275||(3.7)||356||(4.3)||632||(4.0)||564||(5.1)||689||(5.8)||1252||(5.4)|
|191 and 192||Brain and CNS||123||(1.7)||87||(1.0)||210||(1.3)||293||(2.6)||241||(2.0)||534||(2.3)|
|159,165,195–199||All other sites||2193||(29.5)||2781||(33.5)||4974||(31.6)||2906||(26.1)||3517||(29.5)||6423||(27.9)|
The most common causes of cancer death among men between 1994 and 1996 were gastric (20.3%), lung (13.7%), and prostate (12.8%), and in women, gastric (13.9%), cervical (11.7%), and breast carcinomas (9.8%). Cancer of unknown origin (ICD-9 159, 165, and 195–199) accounted for approximately 2000 cases per year (10.3%) in the series, and varied as a proportion of all cancer deaths from 13.3% in 1981–1983, to 9.3% in 1994–1996 (data not shown). Of these, approximately two-thirds corresponded to deaths not certified by a medical doctor.
Time Trends in Mortality Rates
Gastric carcinoma is reported to be the most common cause of cancer death in Colombia, accounting for nearly 20% of the total cancer mortality rate. It is the most common site for both genders, with a male-to-female ratio of 1.3:1. Trends in gastric mortality rates demonstrate a markedly downward trend (Fig. 1) (Table 2), with an average annual decline between 1987–1996 of 2.3% among men and 2.9 % among women (Table 3). The decrease is more apparent among men age < 65 years (representing a 3.1% mean decline per year) compared with older men (1.3% annually).
|Stomach||Colon and rectum||Lung||Cervixa||Prostate|
|EAPC||95% CI||EAPC||95% CI|
|Stomach||−2.3||(−2.3 – −2.1)||−2.9||(−3.0, −2.7)|
|Colon and rectum||+2.6||(2.5 – 2.8)||+1.8||(1.6 – 2.0)|
|Lung||+0.2||(0.1 – 0.3)||+1.3||(1.1 – 1.5)|
|Cervixa||−1.5||(−1.7 – −1.3)|
|Prostate||+3.7||(3.6 – 4.1)|
Colorectal carcinoma ranked the fifth most common cause of cancer mortality rates for both genders. The male-to-female ratio was 0.8. Increasing trends in colorectal carcinoma mortality rates were observed recently for both men and women (Fig. 1) (Table 2). The EAPC for men was 2.6% annually on average, whereas a lower increase of 1.5% was estimated for women (Table 3).
Lung carcinoma is the second most common cause of death among men, and fourth among women. The male-to-female ratio is 1.8. Since the 1980s, mortality trends in lung carcinoma have been increasing rather rapidly among women, but have stabilized among men (Fig. 1) (Table 2). The EAPC for the last 10-year period increased 0.2% and 1.3%, on average, for men and women, respectively (Table 3).
Cervical carcinoma ranked second only to gastric carcinoma among women. Mortality rates began to decline around 1985, but then increased slightly from 1993 onward (Fig. 1) (Table 2). The EAPC between 1987 and 1996 indicates a decrease of −1.5% per year in mortality rates (Table 3), yet a modest increase is observed when only the last 5 years are taken into consideration. The interrelation between the increasing trends in the observed rates of cervical and uterine carcinomas and the decreasing trends in uterine carcinoma unspecified are shown in Figure 2, together with the expected rates of cervical carcinoma on reallocation. After a decline between 1987–1990, cervical carcinoma rates have been increasing.
Although prostate carcinoma was not found to be one of the main causes of cancer mortality in the first years of the series, it ranked as the third most common cause of cancer death overall among men in the last few years. Approximately 85% of all prostate carcinoma deaths occurred among men age ≥ 65 years. The observed trends indicate a constant increase between 1981–1996, with rates doubling from 7.4 in 1981 to 14.4 per 100,000 in 1996 (Fig. 1) (Table 2). Indeed, the mean annual percentage increase of 3.7% between 1987–1996 is the highest increase observed among all sites, confirming that prostate carcinoma is the fastest growing cause of cancer death among Colombian men (Table 3).
The current study describes recent overall trends in the most frequent causes of cancer death in Colombia, a population in which cancer mortality rates have been increasing steadily in the last decades. To our knowledge, the current study is the first report to describe national mortality trends by the main topographic sites. Further work will involve a more detailed age-specific examination of trends as data accrue. Some discussion follows with respect to possible interpretations of the current trends in relation to the known and putative risk factors for the disease, and caveats related to data quality.
Interpretation of Trends
Although trends are in long-term decline, gastric carcinoma mortality is the leading cause of cancer mortality for men and women, with an average annual ASR more than four times higher than that observed in the U.S. during the same period, and very similar to that in Ecuador.8 As in most populations worldwide, trends in gastric carcinoma mortality rates in Colombia have declined markedly over many decades. Although a greater decrease has been reported in the current study for women than for men, other studies have shown larger absolute decreases in death rates for men.9, 10
The decreases are explained, in part, by the reduction in the use of salt as a food preservative, combined with the growing urbanization pattern that favors easy access to fresh food. This, despite the reality that only 54% of Colombian households were in possession of a fridge by 1997.11 The intake of fresh vegetables is still relatively poor in Colombia and a dietary change is unlikely to explain much of the decline in rates. For the period 1965–1970, the Colombian individual supply of vegetables was estimated at an average of 31 kg per year, one-third of the 102 kg per year estimated for the U.S. in the same period. Between 1992 and 1996, vegetable supplies increased only slightly (33.7 kg per year), whereas estimated fruit supplies have decreased from 86 kg per year in 1965–1970 to 40 kg per year in 1992–1996.12
Approximately 40% of gastric carcinoma cases has been attributed to Helicobacter pylori.13 In Colombia, data from a population-based study in the mountain region of the Andes in 199214 estimated the prevalence of H. pylori infection to be as much as 87% among 9-year-old children. Although cohort-related declines in infection have been observed among successive generations in several Western countries,15 no such recent data from Colombia are available.
Mortality rates from carcinomas of the colon and rectum demonstrate a clear upward trend. Although Colombia may be categorized as a population with a relatively low risk for this malignancy, its incidence has increased steadily in recent years, as has been observed in many other developing countries.16, 17 According to data from the Cali Registry in Colombia, incidence rates have increased 3-fold in the last 50 years.18 The observed mortality trends likely reflect the incidence figures. Colombia has no organized early detection program nor is colonoscopy a widely utilized practice. However, there are diverging trends in incidence and mortality in developed countries,19 given that patients with the disease have a relatively good prognosis. The introduction of new chemotherapeutic agents and the early detection of the disease have led to decreases in colorectal mortality rates in a number of countries.9, 20–22
Bearing in mind certain risk factors associated with colon carcinoma—elevated consumption of red meat, heavy smoking, and a sedentary life style23, 24—it could be argued that important changes in lifestyle habits associated with increasing urbanization in Colombia, most likely together with late diagnosis, could explain much of the increasing trend. The stable male-to-female ratio of 0.8 differs from the ratios reported in Europe, in which gender ratios increased during the 1980s, reaching 1.5 by the mid-1990s.25 This has been attributed (in developed countries), in part, to the introduction and increased availability of oral contraceptives,26, 27 as well as to hormone replacement therapy. However, in the U.S., declining trends for both genders, with men having higher rates, suggest that many other factors are involved.28
With respect to lung carcinoma, the slight increase among men, and the more striking increase among women, mostly reflects the different gender-specific patterns of tobacco consumption in previous decades. Historically, higher rates of smoking occur among men than among women in Colombia, but there is evidence of decreasing trends in consumption for both men and women. Among men, smoking prevalence halved since 1979 to 26.8% in 1998. Among women, a similar reduction was observed—smoking prevalence halved since 1979 to 11.3% for 1998, the most recent year.29, 30 Due to the time lag, the reported changes in smoking are not, as yet, reflected in the lung carcinoma trend. It is expected that lung carcinoma mortality rates are plateauing and will start to decrease in the next few years—as has been seen in the U.S. and in several European countries starting in the late 1970s and 1980s.31
Colombian lung carcinoma trends are at odds with most other Latin American countries where increases in smoking prevalence among men and, recently, among women have been observed.32 Increasing mortality trends among women could be attributed partially to errors in death certification, with an overestimatation due to metastatic carcinomas. However, studies that excluded patients ≥ 80 years have found that lung carcinoma death is relatively accurate.16 One hypothesis involves women's exposure to environmental tobacco smoke due to the higher prevalence of smoking among men, given that the increase is larger among older women. Exposure to cooking oil vapors, a putative risk factor among Chinese women, requires further study.33, 34 Industrial exposure to asbestos, mineral dusts, silica, and radon has not to our knowledge been studied in Colombia but studies in Italy35 have found an attributable fraction due to occupational exposure between 20–25%. Ecologic studies36 have estimated attributable risk due to radon exposure to be approximately 3%. Using risk assessment, ≥ 25% of lung carcinomas among nonsmokers > 60 years old could be attributable to radon,37 whereas approximately 5% may be ascribed to radon among smokers.
Cervical carcinoma is the leading cause of death among women in developing countries38 and is the second leading cause of death among women in Colombia. The epidemiologic patterns are of particular concern because death rates have remained consistently high despite the presence of an organized screening program.39 This pattern has also been observed in other Latin American countries.40 Since the establishment of health reforms in Colombia in 1993, the cervical carcinoma screening program has suffered some setbacks, particularly in relation to the difficulty of gaining access to cytology reports. Colposcopy is not covered by the health system available to the poorest people, most likely contributing to the increasing number of observed invasive cervical carcinomas reported. The corresponding recent trends in mortality rates need to be evaluated carefully because they may, in part, serve as indicators of inequity in the provision of health care services.
Some of the upturn may be explained by the corresponding decrease in the rates of unspecified cancer deaths. Clearly, however, some residual increase in the cervical carcinoma mortality rates is left unexplained by the reallocation.
The steady increases observed in prostate carcinoma mortality rates may be a result of detection bias, whereby prostate carcinoma is misclassified as the underlying cause of death in recently diagnosed patients, following the introduction and massive use of the prostate-specific antigen screening test.41 Nevertheless, although this misclassification has declined only since 1994 in the U.S.,17 declines in mortality rates have been observed since 1990 in the U.S. as in other countries.42 This has been related partially to diagnosis at earlier stages, suggesting that late-stage diagnosis is one contributing factor in the prostate carcinoma trends observed in Colombia. In addition, the progressive shift towards urbanization and the adoption of Western lifestyles, particularly a reduction in physical activity and changes in diet,43 also may explain partially the increasing trends.
Data Quality Issues
This examination of trends in death rates, although invaluable, is susceptible to erroneous interpretation, particularly if the quality of the mortality data has itself been changing over time. One important indicator is the quality of death certification, which may cause artefactual increases and decreases in time trends of cancer mortality, and is perhaps even more of a potential hazard to interpretation of trends than underascertainment.44–46
Although the total number of cancer deaths certified by nonmedical practitioners may overestimate cancer deaths in Colombia, these certificates usually correspond to persons living in rural areas without the permanent presence of medical doctors. Therefore, specific subsites may be underestimated, as evidenced by the number of tumors of unknown origin. The current study indicated that 10.7% of cancers were of unknown origin (ICD 195–199). The high percentage of deaths corresponding to liver carcinoma, not specified as primary or secondary (155.2), also highlights another important concern regarding quality, not only in terms of the certification of death, but also in the accuracy of subsite diagnosis. The number of nonspecified uterus deaths is high (28%), but it is within the range of the proportion of unspecified cancers seen in European populations, which is reported to range from 20% to 65%.47
Ultimately, many of the problems related to the quality of the death certification reflect a limited awareness of the importance of mortality data and the potential of these data for use by medical practitioners. Furthermore, the lack of precision of cause of death on the certificate does not permit more detailed analyses, such as trends in colorectal carcinoma mortality rates, although subsites exhibit diverse trends and have a different distribution of risk factors48, 49
To summarize, the demographic and social transition in the Colombian population is the major factor determining increases in cancer incidence and subsequently, increases in cancer deaths. However, the significant increase in cancer mortality rates for several of the most common malignancies also may point to late diagnosis. Further studies to investigate the synergy among variations in cancer incidence, mortality rates, and survival rates in Colombia are needed, as are specific studies on the provision of cancer-related health care services within the new Colombian health care system.
Cancer is now a major public health problem in Colombia. The main causes of cancer death, in spite of the high percentage of tumors of unknown origin, have remained largely unchanged from the distribution observed in the seminal study of Puffer and Griffith of urban area mortality rates.50 It therefore is imperative that cancer control strategies involving surveillance, prevention, education, and research strategies are implemented and developed to prevent unnecessary and premature deaths among current and future generations of Colombian men and women.
The authors are grateful to Dr. D. Maxwell Parkin at IARC for his valuable comments regarding the current study and to Dr. Héctor Posso, from the Instituto Nacional de Cancerología (Bogotá, Columbia) for the support received.
- 1Ministerio de Salud (República de Colombia). Mortalidad y años de vida ajustados por discapacidad como medidas de la carga de enfermedad Colombia 1985-1995 [Mortality and DALY's as a measure of burden of disease, Columbia 1985–1995]. Santafé de Bogotá: Imprenta Nacional, 1999.
- 2Ministerio de Salud (República de Colombia). La carga de la enfermedad en Colombia. [Burden of disease in Columbia]. Bogotá: Editorial Carrera Séptima, 1994.
- 3Departamento Administrativo Nacional de Estadística (Colombia). Censo 93 Resumen Nacional [Population and housing census 1993. National abstract]. Santa fé de Bogota: Departamento Administrativo Nacional de Estadística, 1993.
- 4La mortalidad en Colombia 1953-1991. [Mortality in Columbia 1953–1991]. Santafé de Bogotá: Instituto Nacional de Salud, 1993..
- 5Organización Panamericana de la Salud. Clasificación estadística internacional de enfermedades y problemas relacionadas con la salud, décima revisión. [International classification of diseases, ninth revision]. Publicación Científica. Volume 554. Washington, DC: Centers for Disease Control and Prevention, 1995.
- 6Cancer mortality for selected sites in 24 countries (1950-1957). Tohoku, Japan: Tohoku University of Medicine, 1960..
- 7Stata Corporation. Intercooled Stata for Windows 98/95/NT. 7th edition. College Station, TX: Stata Corporation, 2002.
- 8International Agency for Researchon Cancer. Cancer mondial: worldwide cancer mortality statistics. Lyon: IARC, 2002. Available from URL: http://www-depdb.iarc.fr/who/menu.htm [accessed February 2003].
- 11Departamento Administrativo Nacional de Estadística (Colombia). Boletín especial: encuesta de calidad de vida 1997. Boletín de estadística. Bogotá: DANE, 1998.
- 12Food and Agriculture Organization. Food balance sheets, FAOSTAT 2002. Rome: Food and Agriculture Organization: Available from URL: www.fao.org/faostat [accessed February 2003].
- 13The global health burden of infection associated cancers. Cancer Surv. 1999b; 33: 5–33., , , .
- 16Comparability and quality control in cancer registration. No.19. Lyon: IARC, 1994., , , , , .
- 18Cancer incidence in five continents, volume VII. IARC Sci Pub No. 143. Lyon: IARC, 1997., , , , .
- 24Colorectal cancer. Cancer Surv. 1994; 19-20: 77–98., .
- 29Ministerio de Salud (República de Colombia). III Estudio nacional de salud Bucal II estudio nacional de Factores de riesgo de Enfermedades Crónicas. [Third national oral health study and II national study of risk factors for chronic diseases]. Volume II. Bogotá: Ministerio de Salud (República de Colombia), 1999.
- 30El hábito de fumar en Colombia 1977-1980. [Smoking habits in Columbia 1977–1980]. Bogotá: Editorial Gente Nueva, 1988., .
- 31Trends in cancer incidence and mortality. IARC Scientific Pub. No. 121. Lyon: IARC, 1993., , , , .
- 32Smoking in Latin America: a major public health problem. Cad Saude Publica. 1998; 14 (Suppl. 3): 99–108., .
- 39Instituto Nacional de Cancerología (Colombia). Manual de normas técnicas y procedimientos para la toma de citologías. [Manual of technical norms and procedures for the cytological smear]. Documento técnico.1991-1992. Bogotá: Instituto Nacional de Cancerología, 1992.
- 50Patterns of urban mortalityPan American Health Organization/World Health Organization. Scientific Pub. No. 151. Washington, DC: PAHO/WHO, 1968., .