Monitoring trends in causes of death – India and the Million Death Study

The United Nations Sustainable Development Goals for 2030 include around 15 targets for mortality-related indicators.  The “gold” standard method for monitoring mortality and causes of death is a well-functioning civil registration system in which every death certificate has a medically certified cause of death. Unfortunately, in almost all low-income countries, and some middle income ones, death registration systems do not function well enough to produce data for reliable mortality statistics.  The graph below from our 2017 WHO World Health Statistics report ( shows that around half  of deaths worldwide are registered in a national system with medical certification of cause of death. Not all of these deaths are reported to WHO: about 40 % of global deaths are currently reported to the WHO Mortality database, which collects information on registered deaths and their causes from Member States which have them. Even fewer deaths have a meaningful cause of death code (ICD code), with many in some countries coded to ill-defined causes, symptoms, or non-informative immediate causes such as respiratory failure or heart failure.

The main gaps  in terms of reporting to WHO are China and India, who report only data from their sample registration systems. China has very substantially expanded its sample registration system over the last decade and the per cent of deaths registered with cause of death has increased from 9% in 2005 to over 60% in 2015.

India has a death registration system with medically certified cause of death which essentially captures mostly deaths in urban hospitals, a relatively small proportion of the more than 1 million deaths annually in India. Over the last 10 years, the Indian Registrar General has collaborated since 2001 with the Centre for Global Health Research in Toronto, Canada,  headed by Professor Prabhat Jha, to implement the Million Death Study (MDS) to obtain nationally representative data on causes of death in India. The MDS surveys a national sample of 1.3 million homes within the Sample Registration System (SRS), a continuous demographic surveillance system in operation since 1971.

The MDS uses a modified version of the 2011 WHO verbal autopsy (VA) questionnaire to obtain information on deaths from household members or close associates of people who have died. This questionnaire  uses structured checklist questions about key symptoms and a half-page local language narrative that captures events and their chronology.  Each verbal autopsy is assigned to two of 404 trained physicians, and they classify the underlying cause of death according to the WHO International Classification of Diseases 10th revision (ICD-10) using strict coding guidelines. Differences in diagnosis are reviewed and if agreement is not reached, a senior physician makes the decision (in about 10% of VAs).

Clearly, recall of relevant symptoms may be poor, and some causes of death are much more difficult to distinguish by events and symptoms than others. So VA-based diagnoses tend to have variable and higher uncertainty at the individual level. But in the absence of better data at population level, they do provide useful results for monitoring mortality trends and major causes of death of public health importance.

The Million Deaths Study data are used by my unit at WHO and our academic collaborators as a major input to the estimation of trends and distribution of causes of neonatal, infant and child deaths, as well as the overall age-specific causes of death for India for years 2000-2015, available on the WHO website as part of our Global Health Estimates 2015 (

Last week, the Lancet published a detailed analysis of the MDS results for trends in cause-specific neonatal and 1–59-month child mortality in India from 2000 to 2015 prepared by Prabhat Jha and the Million Death Study Collaborators (who include myself and our key academic collaborators in preparing the WHO-MCEE  and GHE2015 child cause of death estimates). The paper is available at

The Lancet, 380 (2017) . doi:10.1016/S0140-6736(17)32162-1

Among other findings, the study found that for children aged 1-59 months, mortality rates from pneumonia and diarrhoea have fallen by more than 60%. Improved female literacy rates, schemes paying women to deliver babies in hospitals and an increase in spending on public health by the Indian government have all contributed to the falling death rates. Mortality from vaccine-preventable diseases such as measles and tetanus have also fallen substantially. For babies under 1 month of age, mortality rates have fallen but at a slower rate (3.4% per year) than for the 1-59 month age group (5.9% per month). Mortality rates for prematurity or low birthweight have actually risen.

The U.N.’s Sustainable Development Goals seek to end the preventable deaths of newborns and children under five, with countries aiming to reduce infant mortality to at least as low as 25 per 1,000 live births by 2030. To meet the 2030 Sustainable Development Goals for child mortality, India will need to maintain the current trajectory of 1–59-month mortality and accelerate declines in neonatal mortality (to >5% annually) from 2015 onwards. Continued progress in reduction of child mortality due to pneumonia, diarrhoea, malaria, and measles at 1–59 months is feasible. Additional attention to prematurity and low birthweight is required.

Posted in Global health trends, World Health Organization | Tagged , , , , , , | Leave a comment

Sex and gender in reporting health indicators

I am writing a chapter for a US publication, which has tables of statistics for males and females (which I refer to as “sex”).  The editor contacted me to ask whether I should be calling this “gender”.  This reminded me of a Lancet editorial (December 10, 2016) which referred to sex and gender as basic variables in clinical scientific research and requires the reporting of “reporting of sex, gender or both where appropriate”.

Sex is defined in the policy explicitly as biological sex with three categories (male, female, intersex). Gender is explicitly defined as “Gender is a constellation of sociocultural processes that interact with and have the potential to influence human biology.”  In other words, gender is not a variable with a definable set of categories, but a complex web of interacting processes modifying a huge array of health and other personal and group outcomes.  The policy provides no guidance on how it may be possible for researchers to report results by gender.

Because of this confusion of a complex web of interacting processes with a categorical reporting variable, most researchers wanting to report on gender simply use the label “gender” for the biological sex variable collected in their data.

I am by no means dismissing the importance of sex and gender as key determinants of health progress. But the many voices calling for reporting of research by “gender” need to realize that gender is indeed a complex constellation of sociocultural processes, not a set of categories.

Population growth is a major issue for sustainable development, global climate change, etc.  High fertility levels drive population growth, particularly in Africa. The biggest factors impeding reductions in fertility levels relate to high levels of child mortality, lack of access to modern contraception, and gender inequality. In turn, a major determinant for reductions in child mortality is the education of mothers.

All of this relates back to gender inequality, and the inability of women in many societies to have full autonomy in areas relating to sex and reproduction.  We are used to thinking of this as an issue in Africa and the Middle East, associated with premodern cultural and religious traditions, and the Taliban and IS are extreme manifestations of these attitudes.  But similar attitudes exist and are growing in parts of the developed world.  The US right-wing war on women is the obvious example.  Along with strenuous efforts to block access to safe abortion, there are now growing efforts to block access to contraception.  It is very clear that this is not about moral views on the personhood of embryos, it is really about making sexual behavior highly risky as part of the efforts to control the sexuality of women.

Posted in Uncategorized | Tagged , , , | Leave a comment

Global health is improving, so is global health information

Yesterday we released World Health Statistics 2017: Monitoring health for the SDGs. This WHO flagship publication compiles data from the organization’s 194 Member States on 21 health-related SDG targets, providing a snapshot of both gains and threats to the health of the world’s people. While the quality of health data has improved significantly in recent years, many countries still do not routinely collect high-quality data to monitor health-related SDG indicators.

However, there have been improvements in data collection. We highlight that now almost half of all deaths globally are now recorded with a cause. Of the estimated 56 million deaths globally in 2015, 27 million were registered with a cause of death, according to WHO’s annual World Health Statistics. In 2005, only about a third of deaths had a recorded cause. Several countries have made significant strides towards strengthening the data they collect, including China, Turkey and the Islamic Republic of Iran, where 90% of deaths are now recorded with detailed cause-of-death information, compared with 5% in 1999.














As reported last year, overall global health is improving. Average life expectancy at birth has increased by five years globally, and more than 9 years in Africa. This year we report that premature mortality from the four main non-communicable diseases (cardiovascular diseases, cancer, diabetes and chronic respiratory diseases) has declined by 17% since the year 2000.  This is driven mainly by improvements in cardiovascular disease mortality in high income countries, and chronic respiratory disease mortality in low and middle income countries. Cancer mortality is also declining, but more slowly.

The report also includes new data on progress towards universal health coverage. Those data show that globally, ten measures of essential health service coverage have improved since 2000. Coverage of treatment for HIV and bed nets to prevent malaria have increased the most, from very low levels in 2000. Steady increases have also been seen in access to antenatal care and improved sanitation, while gains in routine child immunization coverage from 2000 to 2010 slowed somewhat between 2010 and 2015.

Access to services is just one dimension of universal health coverage; how much people pay out of their own pockets for those services is the other. The most recent data from 117 countries show that an average of 9.3% of people in each country spend more than 10% of their household budget on health care, a level of spending that is likely to expose a household to financial hardship.

A selection of additional results are also available in the press release. This is the 12th edition of World Health Statistics, which has been published annually since 2005.

Posted in Global health trends, Publications, World Health Organization | Tagged , , , , | Leave a comment

University of Sydney Award for contributions to global health

On 27 April, I was awarded the University of Sydney’s President’s Award for my contributions to global health monitoring. I did an Honours Degree in Science majoring in Physics and followed that with a PhD (1979) in theoretical physics. However i then applied my analytic and modelling skills in the field of population health, and my career has seen me working to improve international health outcomes. For 15 years I have led the WHO on global health statistics helping international agencies work together by having access to the same information.

You can read more at the following link:

Posted in Global health trends, World Health Organization | Tagged | Leave a comment

Future life expectancy in 35 industrialized countries: projections to year 2030

Today, the Lancet published our study projecting life expectancy trends to 2030 for 35 countries:

Life expectancy at birth is projected to exceed 90 years for Korean women in 2030, a level of average life expectancy many thought impossible to achieve.  South Korean men are also likely to lead at 84.1 years, closely followed by Switzerland and Australia. The USA is likely to have the lowest life expectancy at birth in 2030 among high-income countries, with levels similar to that of middle-income countries like Croatia and Mexico. My colleague, Professor Majid Ezzati at Imperial College London, says this may be due to a number of factors including a lack of universal affordable access to health services, as well as the highest child and maternal mortality rate, homicide rate and obesity among high-income countries.

The study was led by Majid Ezzati at Imperial College, London, with much of the statistical work carried out by Vasilis Kontis and James Bennett. I collated the country mortality data from the WHO Mortality Database and contributed to the development of the methodology and the interpretation of results. Countries in the study included both high-income countries, such as the USA, Canada, UK, Germany, Australia, and emerging economies such as Poland, Mexico and the Czech Republic. The methods involved development and averaging of an ensemble of 21 Bayesian forecasting models, all of which contributed probabilistically to the final projections. The Bayesian model averaging (BMA) resulted in smaller projection errors than the best single model. The projection errors were assessed by projecting withheld data.

Projected life expectancy at birth in the year 2030

Projected life expectancy at birth in the year 2030

The BMA projections resulted in probability distributions of life expectancy in 2030. The paper included graphical presentations of these posterior probability distributions, as shown in the Figure above  for projected life expectancy at birth. The red dots indicate the median projected life expectancy, and countries are ordered vertically by this median.  A baby girl born in South Korea in 2030 will expect to live 90.8 years (median projection). Life expectancy at birth for South Korean men will be 84.1 years. The British Telegraph newspaper also reproduced one of these graphs, and I suspect it is the first time a tabloid newspaper has given its readers information on Bayesian posterior probability distributions for anything.

The study also calculated how long a 65-year-old person may expect to live in 2030.  The results revealed that the average 65-year-old woman in South Korea in 2030 may live an additional 27.5 years, resulting in an average age at death of 92.5 years.

James Fries in the late 1980s argued that the 85 years represented the upper limit of human life expectancy at population level (the upper limit of individual longevity is somewhere above 120 and probably at present below 130). Others such as Olshansky have also forcefully argued that we will not see much more substantial progress in life expectancy.  However this study suggests we will break the 90-year barrier, and we may still be a long way from the upper limit of life expectancy – if there is one.

I spoke about the results of this study today on the BBC World Service and on the BBC television news. They also published an article on their website:

Other findings from the study include:

  • The five countries with the highest life expectancy at birth for men in 2030 were: South Korea (84.1), Australia (84.0), Switzerland (84.0), Canada (83.9), Netherlands (83.7)
  • The five countries with the highest life expectancy at birth for women in 2030 were: South Korea (90.8), France (88.6), Japan (88.4), Spain (88.1), Switzerland (87.7)
  • The five countries with the highest life expectancy for 65-year-old men in 2030 were: Canada (22.6 additional life years), New Zealand (22.5), Australia (22.2), South Korea (22.0), Ireland (21.7)
  • The five countries with the highest life expectancy for 65-year-old women in 2030 were: South Korea (27.5 additional life years), France (26.1), Japan (25.9), Spain (24.8), Switzerland (24.6)
  • The five countries in Europe with the highest life expectancy at birth for men in 2030 were: Switzerland (84.0), Netherlands (83.7), Spain (83.5), Ireland (83.2) and Norway (83.2)
  • The five countries in Europe with the highest life expectancy at birth for women in 2030 were: France (88.6), Spain (88.1), and Switzerland (87.7), Portugal (87.5) and Slovenia (87.4).
  • The UK’s average life expectancy at birth for women will increase from 82.3 years in 2010 to 85.3 years in 2030. This places them 21st in the table of 35 countries (compared to 22nd in 2010).
  • The average life expectancy of a UK man at birth will increase from 78.3 years in 2010 to 82.5 years in 2030. This places them 14th in the table of 35 countries (compared to 11th in 2010).
Posted in Global health trends, Projections | Tagged , , | Leave a comment

One of the world’s most influential scientific minds…..

I have been included for the second time in the Thomson Reuters list of “Highly Cited erResearchers”. Around 3,000 researchers, in 21 fields of the sciences and social sciences, were selected for the recently released 2016 list based on the number of highly cited papers (in the top 1% for papers in their field for number of citations from January 2004 to December 2014). My papers were classified in the field of medicine, making me one of the 377 “of the world’s most influential scientific minds” in this field, or perhaps more broadly in the health field.  I am the only person from WHO on the list, but several academic colleagues are there:  Harry Campbell (University of Edinburgh), Majid Ezzati (Imperial College), Simon Cousens (London School of Hygiene and Tropical Medicine), Juergen Rehm (University of Toronto) and Chris Murray, Alan Lopez and Mohsen Naghavi (University of Washington).

Posted in Uncategorized | Tagged | Leave a comment

More than 1 billion people globally are living with high blood pressure

This finding come from a new study published last Tuesday in The Lancet, which found that the number of people affected by high blood pressure has almost doubled over the past 40 years:

The study highlighted a stark contrast between where people are most affected, with high-income countries showing declines in blood pressure among their populations in recent decades, while low- and middle-income countries have seen substantial increases — particularly in South Asia and Africa. Half the world’s adults with high blood pressure now live in Asia, and high blood pressure is a condition of poverty, not affluence.

gr4_lrgmgr4_lrgfThe study was led by Professor Majid Ezzati at Imperial College, London in collaboration with WHO staff, including my colleague Gretchen Stevens. Hundreds of scientists around the world also collaborated in compiling data from  1,479 population-based studies that had measured the blood pressures of 19·1 million adults. These data were used to estimate trends from 1975 to 2015 in blood pressure distributions for 200 countries. Other analyses by WHO and the Imperial College group have estimated that raised blood pressure causes  7.5 million deaths globally, almost 13% of all deaths.

This study has provided the most complete picture to date of long-term trends in adult blood pressure for all countries and provides important guidance for addressing the global target set by the WHO World Health Assembly to reduce the prevalence of high blood pressure by 25% by 2025.

Posted in Global health trends, World Health Organization | Tagged , , , | Leave a comment