India Now Has To Choose Between Two Disease Estimates
New Delhi: Did diarrhoeal diseases rank third or seventh in India’s disease-burden ranking for 2017? Where do musculoskeletal diseases such as lower back pain and neck pain stand on this list?
India’s union health ministry may have to answer questions like these now that it has to contend with two different estimates of the country’s disease burden, both computed by the same premier research institute--the Indian Council of Medical Research (ICMR)--but working with two different collaborators.
The answer to the first question is that diarrhoea ranks seventh on the latest disease burden list, the National Burden Estimates (NBE), published by researchers at the National Institute of Medical Statistics (NIMS), the statistical wing of the ICMR. The first-ever NBE was published in the global journal, The Lancet Global Health, on November 8, 2019.
But in the older estimate, diarrhoeal diseases ranked third. This was the 2017 disease-burden ranking by the India state-level disease burden initiative, based on the Global Burden of Disease (GBD) model created by the Seattle-based Institute for Health Metrics and Evaluation (IHME) in collaboration with the ICMR, the think-tank Public Health Foundation of India, and over 100 other organisations. It has been in use since 2017.
Why was a new disease burden estimate introduced? Researchers at the NIMS, as we explain later, could not reproduce the estimates by the GBD and hence decided to find a local, simpler method.
While most NBE and GBD estimates match, there are some critical differences, especially on non-communicable diseases (NCD) and conditions such as nutritional deficiencies. For example, diarrhoeal diseases account for 5.1% of disability-adjusted life years (DALYs)--years of life lost due to ill-health, disability or early death--in the GBD, but only 4.7% in NBE.
A difference of 0.4 percentage points may seem trivial but when converted into premature deaths and lives lived in disability, they become significant for purposes of public health planning and financing.
Also, while the GBD India estimates put the share of NCD in the disease burden list at 55.4% in 2016 (the latest available), NBE puts the figure at 46.6% in 2017, a difference of 8 percentage points. This again is significant to determine how much resource is to be invested in screening and treating NCDs.
The two studies also differ in how diseases are clubbed together: For example, chronic obstructive pulmonary disease (COPD), which results in the inflammation of the airways, is the second highest cause of disease as per the GBD. But in NBE, even after it is clubbed with asthma, COPD is ranked fifth.
Self harm, self-inflicted injury without the intent of suicide, is ranked higher in GBD than NBE, which ranks musculoskeletal diseases (including lower back pain and neck pain) higher.
The NBE study also finds that in rural areas, which account for three-fourths of India’s disease burden, infectious diseases, nutritional diseases, maternal and early infancy diseases dominate. All of these are preventable. Meanwhile, in the urban areas, most diseases are due to NCDs. It is a known fact that as a region develops, its disease burden moves from being predominantly infection-driven to non-infectious, a process described as ‘epidemiological transition’.
|Top 20 Causes Of Disability-Adjusted Life Years (DALYS) In India, 2017|
|Global Burden of Disease, 2017||DALYs (%)||National Burden of Estimates, 2017||DALYs (%)|
|Ischemic heart disease||7.7||Ischemic Heart Disease||9.6|
|Chronic obstructive pulmonary disease||5.1||Perinatal conditions||8.5|
|Diarrheal diseases||5.1||Ill defined||6.3|
|Lower respiratory infections||4.9||Nutritional deficiencies||6.0|
|Neonatal preterm birth||3.8||Chronic Respiratory diseases||5.7|
|Drug-susceptible tuberculosis||2.9||Other Neuropsychiatric conditions||5.6|
|Dietary iron deficiency||2.5||Diarrhoeal diseases||4.7|
|Other neonatal disorders||2.4||Skin, Sense Organ and Oral conditions||4.5|
|Self-harm by other specified means||2.1||Respiratory infections||4.5|
|Intracerebral hemorrhage||1.9||Road Traffic accidents||3.3|
|Diabetes mellitus type 2||1.9||Tuberculosis||3.1|
|Neonatal encephalopathy due to birth asphyxia and trauma||1.8||Liver and alcohol related diseases||3.0|
|Other musculoskeletal disorders||1.7||Musculoskeletal disorders||2.7|
|Falls||1.7||Fever of unknown origin||2.5|
|Low back pain||1.6||Self-inflicted injuries (suicide)||2.4|
|Asthma||1.5||Diabetes, Endocrine and immune disorders||2.3|
|Ischemic stroke||1.3||Other infectious and parasitic diseases||1.9|
|Major depressive disorder||1.2||All other injuries||1.9|
|Age-related and other hearing loss||1.1||Falls||1.8|
Why a new method
The older model, as we mentioned earlier, has been criticised by the NBE as unreliable because its findings could not be replicated by other researchers. “These models [like GBD] are not reliable, or reproducible in part as the ‘black box’ is available only to some researchers based in Seattle [where the IHME is based],” said Prabhat Jha, co-author of the NBE study and professor of global health and epidemiology at the Dalla Lana School of Public Health, University of Toronto.
ICMR, the lead partner for GBD India, countered by arguing that its results are the most widely accepted scientific approach globally. “Easier reproducibility of a particular method does not necessarily mean it leads to more useful findings than a more advanced and scientifically robust approach that requires higher skills,” Balram Bhargava, director general of ICMR and secretary, health research, union health ministry, told IndiaSpend in an email reply.
As NBE has been funded by the union health ministry, IndiaSpend reached out to Nilambuj Sharan, economic advisor, union health ministry and in-charge of the project, via email and phone. We wanted to ask him about the use of the new estimates in ministerial policy decisions. We will update the story when we receive a response.
Other than an ICMR press release dated November 11, 2019, the NBE study has hardly received a mention in the media so far.
Local vs global estimates
The publication of the NBE is an important step forward in the development of health statistics at a national level, said Dinesh Mohan, honorary professor, transportation research & injury prevention programme at the Indian Institute of Technology, Delhi. “It is not a healthy practice for national governments to depend on health data only from the GBD as the government, researchers and policymakers in India have little understanding of [how] those estimates are calculated,” he said.
“One of the issues is that underlying data that the GBD researchers use are of poor quality,” said Denny John, evidence synthesis specialist, Campbell South Asia, New Delhi. Also, NBE is conducted by Indian researchers who are aware of the source and quality issues related to data gathering. GBD is majorly led by researchers based in high-income countries who depend on crowd-sourcing of data from researchers based in assessed countries, John said.
“The issue is not whether one method is better than another; the issue is of transparency,” said a researcher who was involved in the GBD project and had pulled out because of his reservations. He did not wish to be named. Only a handful of people in IHME are aware of the exact models used for regression and this makes assessed countries dependent, he said.
‘Modelling is the most scientific way’
The ICMR has access to all models and methods used to produce GBD estimates, Balram Bhargava said. “There are 15 GBD India expert groups for various diseases and risks, comprising of ICMR scientists and leading experts from India, who have reviewed the models and the findings extensively for the GBD India results,” he said.
The NBE paper’s criticism of modelling is misplaced, said Lalit Dandona, director of the India State-Level Disease Burden Initiative and research professor at the Public Health Foundation of India.
Each data source for disease burden has some bias or the other as a sample of the population is used for the estimates, Dandona pointed out. GBD uses multiple data sources of all the available evidence to arrive at estimates that are closer to the true disease burden at the population level than is possible using a single or a few select sources of data.
“The GBD modelling approach is informed by a quarter century of scientific advances,” Bhargava wrote in an email and added that the World Health Organization (WHO) has now entered a collaboration with the GBD to publish disease burden estimates.
History of the estimates
The India State-level Disease Burden Initiative, released in 2017, provided findings for distribution of diseases and risk factors from 1990 to 2016 for all states. The effort took two years and the expertise of 200 experts from 102 organisations.
One of the key findings of the report was the rising threat of non-communicable diseases that killed 61.8% of people in 2016 as compared to 37.9% in 1990, as IndiaSpend reported in November 2017.
The GBD method used models to analyse over 90,000 data sources including census, vital registration, disease registrations and other surveys to produce estimates for death and disability from various diseases.
The NBE project was initiated in 2017 with the support of the health and family welfare ministry to provide “transparent and understandable” estimates at the national and state level. It used open-source data from the United Nations, the WHO and the Million Death Study (MDS)--a survey to determine the causes of death in India where most deaths go unregistered.
The publishing of NBE was a landmark event for India because it employed transparent methods and its results are reproducible, Geetha Menon, scientist, ICMR-NIMS, and lead author of the paper in The Lancet, told IndiaSpend.
“It provides a simple, locally operable tool to aid policy makers in priority setting in India and other low-income and middle-income countries,” said the NBE paper.
As we said, there is more than an 8-percentage-point difference between NCD estimations by NBE and GBD.
“This is because almost 14.8% of all the deaths in India has been categorised as ill-defined,” said Menon. While NBE has created a category for ill-defined deaths, the GBD redistributes them in other categories.
The funds for NBE totalled Rs 2.47 crore of which Rs 14 lakh were used for analytics. The NBE team has also created an interactive tool which can help health researchers calculate the disease burden at different population levels, and can be used even at the district level.
The NBE has also published a report for 2015 which has been submitted to the union health ministry.
The need for a national estimate
In July 2018, the WHO and IHME signed a new memorandum of understanding to strengthen collaboration on GBD. This means that from 2019, there will be a single disease burden study instead of two separate ones by the WHO and the IHME.
There have been other critiques of GBD estimates, especially mortality estimates which often differ from actual estimates. For example, the ranking and figures for child mortality in GBD were very different from those from Eurostat, the agency that provides statistics for the European Union, as this letter to The Lancet pointed out in January 2019.
Even when the South African Medical Research Council, a premier research body, published its second National Burden of Disease estimates in 2016, it found that GBD’s HIV AIDS estimates were much higher than indicated in the national registry. It also found other variations in mortality figures.
“We caution policy makers not to use country-specific data from the GBD blindly, and urge the GBD collaborators to pay more attention to estimates produced by organisations and governments within countries,” South African researchers wrote in The Lancet.
(Yadavar is a special correspondent with IndiaSpend/HealthCheck.)
We welcome feedback. Please write to firstname.lastname@example.org. We reserve the right to edit responses for language and grammar.