Country-level correlates of cervical cancer mortality in Latin America and the Caribbean

Cervical cancer in Latin America Artículo original Country-level correlates of cervical cancer mortality in Latin America and the Caribbean Ana Pereira-Scalabrino, MD, PhD,(1) Maribel Almonte, PhD,(2) Isabel dos-Santos-Silva, MD,PhD.(3) Pereira-Scalabrino A, Almonte M, dos-Santos-Silva I. Country-level correlates of cervical cancer mortality in Latin America and the Caribbean. Salud Publica Mex 2013;55:5-15. Pereira-Scalabrino A, Almonte M, dos-Santos-Silva I. Determinantes a nivel país de la mortalidad por cáncer cervicouterino en Latinoamérica y el Caribe. Salud Publica Mex 2013;55:5-15. Abstract Objective. To identify country-level correlates of geographical variations in cervical cancer (CC) mortality in Latin America and the Caribbean (LAC). Materials and methods. CC mortality rates for LAC countries (n=26) were examined in relation to country-specific socio-economic indicators (n=58) and Human Papilloma Virus (HPV) prevalence using linear regression models. Results. High mortality at ages <5 years, low per capita total expenditure on health, and low proportion of the population with access to sanitation were identified as the best independent predictors of CC mortality (R2=77%). In the subset of countries (n=10) with HPV prevalence estimates, these socio-economic indicators together with high-risk HPV prevalence explained almost all the between-country variability in CC mortality (R2=98%). Conclusion. The findings suggest that continuing socioeconomic improvements in LAC countries will be associated with further reductions in CC mortality even in the absence of organised population-based screening and vaccination programmes. Resumen Objetivo. Identificar variables a nivel de país que expliquen las variaciones geográficas en la mortalidad por cáncer cervicouterino (CaCu) en América Latina y el Caribe (AL). Materiales y métodos. Se examinaron las tasas de mortalidad por CaCu de cada país (n=26) mediante modelos de regresión lineal en relación con indicadores socioeconómicos (n=58) y prevalencia del virus del papiloma humano (VPH). Resultados. Alta mortalidad en menores de cinco años, bajo gasto total en salud per-cápita y baja proporción de población con acceso a saneamiento básico son los mejores predictores de mortalidad por CaCu (R2=77%). En los países (n=10) con estimaciones de prevalencia de VPH, estos indicadores socioeconómicos y la prevalencia de VPH de alto riesgo explicaron el 98% de la variabilidad de CaCu en AL. Conclusión. Las mejoras en el nivel socioeconómico en AL están asociadas con reducciones en la mortalidad por CaCu, a pesar de la ausencia de programas organizados de tamizaje e inmunización contra VPH. Keywords: uterine cervical neoplasms; papillomaviridae; mass screening; Latin America Palabras clave: neoplasias del cuello uterino; papillomaviridae humano; tamizaje masivo; América Latina (1) (2) (3) Unidad de Epidemiología Nutricional y Genética, Instituto de Nutrición y Tecnología de los Alimentos, Universidad de Chile. Santiago, Chile. Cochrane Institute of Primary Care and Public Health, Cardiff University. Wales, UK. Department of Non-Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine. England. Received on: February 21, 2012 Accepted on: June 19, 2012 Corresponding author: Ana Pereira Scalabrino. Unidad de Epidemiología Nutricional y Genética, Instituto de Nutrición y Tecnología de los Alimentos, Universidad de Chile. Av. El Líbano 5524. Macul, Santiago, Chile. E-mail: salud pública de méxico / vol. 55, no. 1, enero-febrero de 2013 5 Pereira-Scalabrino A y col. Artículo original C ervical cancer (CC) is the second most common female non-skin cancer in the world. According to Globocan 2008,1 13% of all CC cases and 11% of all CC deaths worldwide occur in Latin America and the Caribbean (LAC), with rates being higher than in more developed countries. There is, however, marked between-country variability in CC rates in LAC with a four to five-fold difference in rates between high (e.g. Nicaragua and Guyana) and low risk countries (e.g. Uruguay and Chile).1,2 Persistent infection with highrisk Human Papillomavirus (HPV) is a necessary cause for CC development,3 but geographical variations in the prevalence of HPV do not seem to fully explain the variability in CC rates worldwide.4 Organised cervical screening programmes based on cytology can reduce CC incidence and mortality rates5 by as much as 8090%.6 Both ecological and individual-based studies have shown that markers of socio-economic (SE) status, such as educational level, are related to CC incidence and mortality, with women with low SE status being at higher risk of developing, or dying from, this cancer.7,8 We conducted an ecological study to assess the extent to which between-country differences in CC mortality in LAC are accounted by level of SE development, HPV prevalence and screening activity. Materials and methods CC mortality rates Country-specific CC mortality rates (age-adjusted to the World standard population) for 2008 were extracted from GLOBOCAN 2008.1 These rates were estimated from death certification data provided by each country to the World Health Organization (WHO) except for Bolivia, Guyana, Honduras, Jamaica and Haiti. For these countries, CC mortality estimates were corrected for under-reporting (Guyana), or derived using data on CC incidence rates and survival (Bolivia) or CC mortality from neighbouring countries (Honduras, Jamaica).1 Haiti was excluded from the analysis because its GLOBOCAN estimates were markedly different for 2002 and 2008 (i.e. age-adjusted rates of 48 and 10 per 100 000 women, respectively).1,9 Level of SE development Data on demographic, SE and public health indicators for the years 2000-2005 were extracted, for each LAC country, from official web pages or reports published by non-governmental organisations, including the WHO,10 the Pan-American Health Organisation (PAHO),11-13 the United Nations Statistics Division14 6 and Human Development Report.15 These indicators were categorised into nine strata (table I): demographic (eight indicators), mortality (eight), morbidity (two), immunisation coverage (five), tobacco use (two), sexual and reproductive behaviour (six), health services (ten), economic (eleven) and development (six) indicators. Linear univariate regression models, weighted by size of the female population in each country, were fitted to examine the association between each one of these 58 indicators and CC mortality at a country level. For each one of the nine strata described above, the indicator with the highest R2, a p-value<0.05 and available data for all LAC countries examined was chosen to be included in a multiple regression model. The tobacco use stratum was excluded because none of its indicators had information for all the countries examined (table I). The correlation between the remaining eight selected stratum-specific indicators was then evaluated and whenever two (...truncated)