PROGRESS OF CGHR FROM 2003-2006

The Centre for Global Health Research (CGHR) was established in 2003 to conduct large scale epidemiological studies in developing countries. The CGHR is co-sponsored by St. Michael’s Hospital and the University of Toronto. It has offices in Toronto, New Delhi and Bangalore. CGHR is affiliated also with the Centre for International Health, the McLaughlin Centre for Molecular Medicine and other partners at the University of Toronto.

 A. SUMMARY OF RESULTS AND CONCLUSIONS OF RECENT WORK                                     The chief research focus of the Centre for Global Health Research (CGHR) is the epidemiology and prevention of premature mortality in developing countries, more specifically the measurement of the epidemiological correlates (such as smoking, alcohol, blood pressure, blood lipids, diabetes, obesity, and HIV-1 related risk behaviors) of premature adult mortality. This involves mega-scale population-based studies that use “large, simple” prospective and retrospective study designs and methods to reliably document outcomes and exposure.The following results and conclusions have been achieved.

1. Demonstrating the feasibility of mega-sample cohort studies of risk factors for premature mortality in developing countries (“Prospective Study of 1 Million Deaths in India). Working with the Indian government and global collaborators, we have designed and implemented two prospective studies involving nearly 14 million people in 2.4 million nationally representative Indian households (6.3 million people in 1.1 million households in the 1998–2003 sample frame and 7.6 million people in 1.3 million households in the 2004–2014 sample frame (Jha et al, 2006a). Households are monitored for vital status and, if dead, the causes of death through verbal autopsy. About 300,000 deaths from 1998-2003 and some 700,000 deaths from 2004-2014 are expected; of these about 850,000 will be coded by two physicians to provide causes of death by gender, age, socioeconomic status, and region. These prospective studies will document reliably the established risk factors for premature mortality by monitoring the development of disease in those with or without these risk factors. The studies have several innovations that are relevant to other developing countries considering the measurement of mortality and to recent calls for improved global health statistics (Lancet editorial, 2005). It uses an innovative household instrument to ascertain causes of death and dual recording methods to improve reliability and consistency. It is national in scale, representative of the population, and – by recording underlying demographics – able to quantify absolute mortality rates. Our planned addition of biological samples will enable the examination of genetic and biological correlates of disease. Pilot study among 9000 is completed in  December 2006 and analysis is ongoing.

2. Demonstrating the wide-scale feasibility and validity of “verbal autopsy” to documentcauses of death. India and other developing countries urgently need the development of health information systems adapted to the current and future problems in public health. This includes reliable quantification of the causes of death: over 75% of the annual estimated 9.5 million deaths in India occur in the home without medical attention. Thus the large majority of deaths do not have a certified cause. In these circumstances, we have developed a widely practicable “verbal autopsy” tool to document causes of death. Preliminary results from over 35,000 deaths in the prospective study (plus some 80,00 deaths from earlier retrospective studies) suggest that verbal autopsy can ascertain the leading causes of death, minimize the misclassification of causes, and derive the probable underlying cause of death when it has not been reported. Agreement on cause-of death patterns between two independent teams is high. Verbal autopsy yields broad classification of the underlying causes in about 90% of deaths before age 70. In old age, however, the proportion of classifiable deaths is lower (Jha et al, 2006a).

 3. Reliable large-scale measurement of existing smoking, alcohol and other “established” risk factors. Our baseline study of 1.1 million households surveyed in 1998 in India has already documented marked variation in male smoking and alcohol drinking (few females report smoking or drinking; Jacob et al, 2006; RGI, 2005). There is ten-fold variation from one state to another in smoking, as well as marked variation by education and socioeconomic background.

Similarly, we have been able to document variation in living conditions (type of housing, water and sanitation access, indoor air pollution), as well as in fertility practices- both of which are particularly relevant for child survival. We have documented, for example, that observed second or third female births following earlier female births are 30-40% less common than expected (Jha et al, 2006b).

The difference is explicable only by use of prenatal sex determination and selective abortion of female foetuses, which we estimate, conservatively, to account for about 0.5 missing female births a year (or about 10 million since the widespread availability of ultrasound over two decades). This study has received worldwide attention.

 4. Development of innovative retrospective methods to study risk factors. In addition to prospective methods, determinants of death can be identified by comparing risk factors between the dead and living. Such household case-control studies use the dead as cases and their surviving spouses or close relatives as controls. A retrospective study in Chennai of 43,000 male deaths and 35,000 living controls (Gajalakshmi et al, 2003), using these methods, has documented that throughout middle age, the death rates from medical causes of smokers were double of those of non-smokers (standardized risk ratio at ages 25–69 of 2.1, with 95% confidence interval 2.0–2.2, smoking-attributable  fraction 31%). A large part of this excess risk was from tuberculosis and vascular deaths. If these hazards are similar across India, then about half of all tuberculosis deaths in India could be accounted for by smoking. The same study also suggests that smoking is more likely to spread tuberculosis, by converting sub-clinical to clinical diseases. Similarly, we have used “proportional mortality” methods to study risk factors. Preliminary studies among childhood deaths in Northern India find that 61% of children who died of vaccine-preventable diseases were not immunized in comparison to 40% of control children who died from injuries. The crude odds ratio of 2.4 suggests that half of the vaccine-preventable child deaths need not have occurred. Using these two retrospective methods, we will examine immunization, childhood malnutrition, alcohol, male time away from home (as a proxy of HIV-1-related sexual risk taking), and other variables.

5. Documenting HIV-1 spread in developing countries. We have been actively researching the causes of HIV-1 spread. This includes systematic reviews of interventions to reduce HIV-1 transmission (Jha et al, 2001), mathematical modelling and trend projection in Botswana and India (Nagelkerke et al, 2002), meta-analyses of 173 African and 6 Indian epidemiological studies (Chen et al, 2006), quantifying correlates of infection among 32,000 attendees at voluntary counselling and testing centres in Tamil Nadu, ecological correlates of HIV-1 in 115 districts (Pupp et al, 2006) and documenting trends of HIV-1 among 424,000 women in India from 1998-2004 (Kumar et al, 2006).

 6. Developing evidence based course on HIV/AIDS programming in developing countries. Centre for Global Health Research (CGHR) utilizing consultation with international advisors from UNAIDS, WHO, and the Global Fund for HIV, TB and Malaria, with support from the International  AIDS Society, developed and successfully implemented ‘The Short Course on Evidence-based HIV Control for Program Managers from Developing Countries’ from Aug. 19-24, 2006, immediately following the International AIDS Conference in Toronto. The program consisted of five days of intensive learning on current issues of HIV/AIDS in developing countries, the evidence available and its policy implications.

The course brought together a group of 30 participants consisting of policymakers, government officials and program managers from 21 countries in Asia, Africa and the Caribbean’s. This event provided a unique platform for future networking for global knowledge exchange and supportive interaction among participants and faculty.

7. Ensuring the relevance of epidemiology to control programs worldwide. A substantial priority of our work is to ensure that research evidence is used and understood by program officials in developing countries and in international institutions. Prabhat Jha’s earlier work on tobacco epidemiology  (paired with economics) led to two widely influential reports on tobacco control (Jha & Chaloupka, 1999; 2000a, 2000b). These reports formed the technical basis into the WHO’s Framework Convention on Tobacco Control, a worldwide treaty to reduce tobacco use. Similarly, CGHR’s HIV-1 epidemiological studies have helped demonstrate the effectiveness and widespread practicability of peer-based education and condom programs to interrupt HIV-1 transmission from female sex workers to male clients. Such strategies formed the basis for the World Bank’s $200 M control program (1999-2005), and the Bill and Melinda Gates Foundation program (2004-2008) in India and in other countries. Partly as a result, HIV-1 prevalence among young women in South India (but not, as of yet, North India) has fallen by about 40% since 2000 (Kumar et al, 2006). Prabhat Jha led the analytic work for the WHO’s Commission on Macroeconomics and Health (Jha et al, 2002b) and is a co-editor of the Disease Control Priorities Project (Jamison et al, 2006; www.nih.gov/fic/dcpp/), which will synthesize  evidence on the effectiveness of interventions across some 80 conditions and risk factors to reduce premature mortality.