Prabhat Jha

Centre for Global Health Research (CGHR) St. Michael’s Hospital and Dalla Lana School of Public

Health, University of Toronto

Prabhat.jha@utoronto.ca Twitter: Countthedead

Asian Development Bank, May 13, 2014

Avoidable Mortality in Urban and Rural India

Disclaimer: The views expressed in this paper/presentation are the views of the author and do not necessarily reflect the views or policies of the Asian Development Bank (ADB), or its Board of Governors, or the governments they represent. ADB does not guarantee the accuracy of the data included in this paper and accepts no responsibility for any consequence of their use. Terminology used may not necessarily be consistent with ADB official terms.

Conclusions

1. Counting the dead and describing causes are central to reducing premature mortality in urban and rural India in the 21st century

2. Three low-cost, high-impact applications of science relevant to urban and rural India:

- Child survival - Malaria control - Tobacco control

0

20

40

60

80

100

0 10 20 30 40 50 60 70 80 90 100

Males, England & Wales,

% survival at period rates

Age

%

Source: Gary Whitlock, CTSU from Registrar-General reports and Human Mortality Database * Males and females combined- from Edmond Halley, 1693 for Breslaw, Germany

2010

1960

1910

1860

Germany-1700*

0

20

40

60

80

100

0 10 20 30 40 50 60 70 80 90 100

Males, England & Wales,

% survival at period rates

Age

%

Source: Gary Whitlock, CTSU from Registrar-General reports and Human Mortality Database

2010

1960

1910

0

20

40

60

80

100

0 10 20 30 40 50 60 70 80 90 100

Males, England & Wales,

% survival at period rates

Age

%

Source: Gary Whitlock, CTSU from Registrar-General reports and Human Mortality Database

2010

1960

0

20

40

60

80

100

0 10 20 30 40 50 60 70 80 90 100

1960 China, India, Ethiopia, versus

2010 England

Age

%

Source: Untied Nations

England 2010

China

India

Ethiopia

0

20

40

60

80

100

0 10 20 30 40 50 60 70 80 90 100

2010 China, India, Ethiopia, versus

2010 England

Age

%

Source: United Nations

England

China

India

Ethiopia

~1.5 M at ages 0-4

Age at death Childhood/early adulthood* age 0-29 2.6 M Middle age age 30-69 4.2 M Old age age 70+ 3.0 M TOTAL 9.8 M

Deaths in India, 2012

Age at death 2012 Future deaths deaths (M) at current death rates Childhood/early adulthood* age 0-29 2.6 ~3 Middle age age 30-69 4.2 ~9 Old age age 70+ 3.0 ~14 TOTAL 9.8 26

Future premature deaths in India

~10 M Deaths ~26 M Births

“for sanitary purposes it is indispensable

to know the relative mortality in small

and, as far as possible, well-defined

tracts to ascertain the death rates in each

of these communities; to see how far this

arises from preventable causes; and to

apply the remedies”

Sanitary Commissioner of the

Government of India, 1869

• Nationally representative sample (Sample Registration System)

• ~7900 of these small areas randomly chosen from all parts of India (each with about 1000 people per area)

How was the study done?

MILLION DEATH STUDY IN INDIA 1. Visit 1 M homes (“true snapshot” of India) with a recent death &

ask standard questions and get a narrative 2. Use non-medical surveyors (electronic entry + GPS) 3. Web-based double coding by 500 doctors (guidelines +

adjudication and other strict quality control) 4. Study all diseases, work with census dept, keep costs <$1 per

home

MILLION DEATH STUDY: selected results (M=Millions, K=thousands)

•4-12M girls aborted before birth since 1980 (1/2 of these since 2000)

•1M smoking deaths (more than expected) and 0.1M alcohol deaths

• 200K malaria deaths: WHO predicted only 15K

• 100K HIV deaths: UNAIDS predicted 400K

• 60K pedestrian traffic deaths: Police estimate=9K

• 50K snakebite: WHO worldwide estimate=50K

• 33K cervical cancer: only 7K at Kashmir/Assam rate

• Each common disease is rare somewhere in India, & hence is largely avoidable

The big avoidable causes of deaths

per year, India, M 2005-2010 • Neonatal causes: 1.0

• Childhood ages 1-59 months 0.8

• Tuberculosis all ages 0.4

• HIV ages 15-59 0.1

• Malaria ages 1 month-69 years 0.2

• Heart attacks ages 30-69 0.7

• Stroke ages 30-69 0.3

• Cancer ages 30-69 0.4

Different MDS results of rural and urban ages 5-69

years

Disease Rural Urban

Malaria 3.6 2.1 P<0.001

Tuberculosis 9.4 7.5 NS

HIV/STI 0.7 0.7 NS

Other infectious diseases1 14.1 8.8 P<0.01

Maternal conditions 1.8 0.9 P<0.01

Cancer 8.3 10.6 P<0.01

Ischemic heart disease 10.6 18.1 P<0.01

Stroke 6.9 8.4 P<0.01

Chronic resp. disease 9.1 6.9 NS

Liver cirrhosis 3.6 5.4 P<0.01

Renal and other endocrine diseases 1.7 2.8 P<0.01

Road traffic accidents 3.0 4.6 NS

Suicides 4.4 3.2 P<0.01

Ill-defined 5.5 3.9 P<0.01

Aleksandrowicz et al 2014 BMC Med

Rank order OR: hospital vs. home Disease Home Hospital

Malaria 3.7% 3.0% NS Tuberculosis 10.8% 5.9% p<0.001 HIV/STI 0.8% 0.5% p<0.001 Other infectious diseases1 14.6% 10.7% p<0.001 Maternal conditions 1.2% 3.2% p<0.001 Cancer 9.5% 9.1% NS Ischemic heart disease 11.3% 14.8% P<0.05 Stroke 7.5% 8.4% P<0.05 Chronic resp. disease 10.6% 4.8% P<0.001 Liver cirrhosis 3.9% 5.1% NS Other digestive diseases 2.3% 2.6% NS Renal 1.9% 2.6% P<0.01 Road traffic accidents 0.5% 4.5% P<0.001 Ill-defined 5.9% 3.4% p<0.001

Different MDS results of home vs hospital deaths

ages 5-69 years

Aleksandrowicz et al 2014 BMC Med

Under-5 mortality progress 2001-2012

100%

100%

89%

80%

74%

70%

64%

58%

54%

52%

46%

43%

15%

13%

6%

5%

3%

2%

0%

0%

Tamil Nadu

Kerala

Maharashtra

Punjab

West Bengal

Karnataka

Jammu & Kashmir

Himachal Pradesh

Uttarakhand

Haryana

Gujarat

Andhra Pradesh

Assam

Jharkhand

Chhattisgarh

Bihar

Rajasthan

Madhya Pradesh

Uttar Pradesh

Orissa

Disrticts on track to meet under-5 mortality target for 2015 Districts

On track / Total

32 / 32

14 / 14

31 / 35

16 / 20

14 / 19

21 / 30

14 / 22

7 / 12

7 / 13

11 / 21

12 / 26

10 / 23

4 / 27

3 / 24

1 / 18

2 / 38

1 / 33

1 / 50

0 / 71

0 / 30

Neonatal & 1-59 month mortality progress

2001-2012

81 districts are home to 37% of the national deaths

in children < 5 years

68 of these 81 districts are in poorer states

Girl disadvantage in 1-59 month mortality

• Nationally: for every 100

boys who died at 1-59

months, 131 girls died.

• Female mortality at these

ages exceeds male

mortality by more than 25%

in 303 districts

• Excess female mortality is

seen in nearly all states

including Kerala and Tamil Nadu

• Nationally: about 74 000

excess deaths in girls at these ages

Age at death 2012 Future deaths deaths (M) at current death rates Childhood/early adulthood* age 0-29 2.6 ~3 Middle age age 30-69 4.2 ~9 Old age age 70+ 3.0 ~14 TOTAL 9.8 26

Future premature deaths in India

~10 M Deaths ~26 M Births

Cumulative risk of death between ages 0-14 by gender, 2012

Cumulative risk of death between ages 15-69 by gender, 2012

Ram et al 2014 in press

Mean height of men (15-54) % anemia in men (15-54) Malaria rates 2005/6 2005/6 1948

Ram et al 2014 in press

The big avoidable causes of deaths

per year, India, M 2005-2010 • Neonatal causes: 1.0

• Childhood ages 1-59 months 0.8

• Tuberculosis all ages 0.4

• HIV ages 15-59 0.1

• Malaria ages 1 month-69 years 0.2

• Heart attacks ages 30-69 0.7

• Stroke ages 30-69 0.3

• Cancer ages 30-69 0.4

0

10

20

30

40

50

60

70

80

Dea

thra

tepe

r1

00

00

0

Age range

0 − 4 5 − 14 15 − 29 30 − 44 45 − 59 60 − 69

591

349

388

319

500

538

Age-specific India malaria-attributed death

rates estimated from the MDS and those

estimated indirectly for WHO

WHO indirect estimates of Indian malaria mortality rates

MDS-attributed Indian malaria mortality rates

Source: Dhingra, et al; Lancet Oct 2010

Geographic distributions of malaria-attributed

mortality and slide P. falciparum rate

0 − 0.75%

0.75 − 1.5%

1.5 − 2.5%

2.5 − 5%

over 5%

Study-attributed malaria mortality

as percent of all mortality

at ages 1 month to 70 years

a

Slide P. falciparum rate 1995-2005

derived from the National Vector-borne

Disease Control Programme

b

0 − 0.58

0.58 − 0.81

0.81 − 1.14

1.14 − 1.53

over 1.53

High-malaria states

ORCG

JH

NE

ORCG

JH

NE

Source: Dhingra, et al; Lancet Oct 2010

MEN

• Oral 45,800 or 23%

• Stomach 25,200 or 13%

• Lung 22,900 or 11%

WOMEN

• Cervical 33,400 or 17%

• Stomach 27,500 or 14%

• Breast 19,900 or 10%

Leading cancers in men and

women, age 30-69 years

Source: Dikshit et al, Lancet 2012

MEN WOMEN

INDIA 47 44

Rural 46 45

Urban 49 42

Richer state 51 46

Poorer state 39 38 Source: Dikshit et al, Lancet 2012

Deaths among 1000 30 year

olds before age 70 from cancer,

at 2010 death rates

Cancer death rates by

education, men and women aged 30-69, India

107

93

46

107

64

43

0

20

40

60

80

100

120

Illiterate Primary Above secondary

Educational level

Ag

e-s

tan

dard

ised

death

rate

(p

er

100,0

00)

Men

Women

Source: Dikshit et al, Lancet 2012

Cancer (non tobacco/non infection): BOTH GENDERS aged 30-69 years

Source: Dikshit et al, Lancet 2012

MEN • Ischemic heart: 0.45 M

- 0.37 M had prior history

• Stroke: 0.20 M

WOMEN

• Ischemic heart: 0.20 M

- 0.16 M had prior history

• Stroke: 0.15 M Deaths from: heart failure (50,000), Rheumatic (10,000)

INDIA: 1 M vascular disease

deaths at ages 30-69 years

Source: Gupta et al, forthcoming

Vascular mortality by level of education,

ages 30-69 years

Source: Gupta a et al, in press

Patients already diagnosed with a stroke

or heart attack (MI): prevent recurrence

by combining 3-4 generics in 1 daily pill

Randomised Trial result: annual comparison rate of stroke/MI

Aspirin vs nothing 5% vs 7%

Aspirin + (BP lowering vs not) 3% vs 5%

BP lowering + aspirin + (statin vs not) 2% vs 3%

10-year risk: 50% if untreated vs 16% with 3 drugs

For every 20M on treatment, prevent 1M events / yr

Post heart-attack mortality

by income, Indian males Income/

death rate Rich

Upper

middle

Lower

middle Poor P

Death rate

(unadjusted) 5.5 5.9 6.5 8.2 <0.0001

Death rate

(RR; adjust for risk factors)

5.1

1.0

5.9

1.3

6.7

1.3

7.8

1.6

0.0093

Death rate

(RR; adjust for risk factors and treatment)

6.9

1.0

7.0

1.0

6.5

0.9

6.7

1.0

0.949

Risk factors: age, sex, prev MI, DM, HTN, smoking, BP, HR, Killip, BMI

Treatments: type of hospital, time to hosp, in-hospital drugs, interventions.

Xavier et al, Lancet 2010

Worldwide no of substance users B=billions, M=millions

Substance Users Annual deaths Smoking 1.3 B * ~ 5 M

Drinking 2.0 B ~ 2 M

Illicit drugs 0.2 B ~ 0.2 M

Global sales ~6,000 B sticks (vs 5,000 B in 1990)

1 ton of tobacco=1 M sticks=1 death

Source: WHO 2002

40 50 60 70 80 90 100

Yearly dots

BMI, kg/m2

30-35 (~32) 40-50 (~43)

0

20

40

60

80

100

Age (years)

% s

urv

iva

l fr

om

ag

e 3

5

Never- smokers

Cigarette smokers

Prospective Studies Collaboration (males)

0

20

40

60

80

100

40 50 60 70 80 90 100

Male British Doctors’ Study

Yearly dots

Low-mortality BMI

Severe obesity

22½-25 (~24)

10 years

Moderate obesity

Obesity and smoking: life expectancy Lose 3 years with moderate obesity/10 years with smoking 2 kg/m2 extra BMI (if overweight) or 10% smoking prevalence shortens life by ~1 yr

Source: Peto, Whitlock, Jha, NEJM, 2010

Source: Lancet 27 June 2009

Russian male death rate ratios

~1 bottle of vodka/day

vs <0.5 bottles/week:

2 x any medical cause

4 x road traffic accident

6 x any other accident

8 x suicide

10 x murder

Source: Lancet 27 June 2009

Survey US women and men & link them to the National Death Index “Facebook of death” (Hazard ratios* current vs. never smokers,

ages 25-79, by gender)

WOMEN WHO SMOKE: 3.0 times more likely to die MEN WHO SMOKE : 2.8 times more likely to die

Source: Jha et al, NEJM, Jan 24, 2013

FEMALES: Survival probabilities between ages 25 and 80 years among current and never-smokers in the US

HR adjusted for age, education, alcohol,

adiposity (BMI), scaled to 2004

national rates, but comparable results if only actual cohort

used

Source: Jha et al, NEJM, Jan 24, 2013

MALES: Survival probabilities between ages 25 and 80 years among current and never-smokers in the US

Source: Jha et al, NEJM, Jan 24, 2013

HR adjusted for age, education, alcohol, adiposity (BMI), scaled to 2004 national rates, but comparable results if only actual cohort used

Source: Jha et al, NEJM, Jan 24, 2013

10

9

6

4

25-34

35-44

45-54

55-64

Years gained by quitting smoking by age

Delayed hazard: observed (1950, 1990) and predicted (2030) proportions of all deaths at

ages 35-69 due to tobacco

US (all adults) China (men)

1950 12% 1990 12%

1990 33% 2030 33%

Source: Peto et al, Nature Medicine, 2001

Chinese cigarette increase 40 years after US increase

Men who smoke bidis 6 years

Women who smoke bidis 8 years

Men who smoke cigarettes 10 years

INDIA: Years of life lost among 30 year old smokers* (MDS results)

* At current risks of death versus non-smokers, adjusted for age, alcohol use and education

(note that currently, few females smoke cigarettes)

Source: Jha et al, NEJM, Feb 2009

Cigarette prices

tripled, smoking

halved, revenue

doubled:

FRANCE and

SOUTH AFRICA

Jha and Peto NEJM 2014

UK & France, lung cancer mortality trends (35-44) to

1997, but not beyond

Male

Male

Source: Peto, 2012

What can be done?

• Much faster progress in reducing under-5 mortality is

needed in all districts

• Focus resources on the 81 districts accounting for 37%

of all under-5 deaths

• More districts lag behind the neonatal goals than the

1–59 month mortality goals in both richer and poorer

states, suggesting a need for renewed national

attention on strategies to reduce neonatal deaths

• All districts could benefit from better accountability

and assessment of their performance, including the

causes of neonatal death

A blueprint for better health

Recent DCPN report

Burden of disease

Disease control priorities

Health system reform

Policy relevance

Increased health spending

Decentralization

District Report Card:

Comparative Performance

CGHR.ORG

Specific actions for each district

CGHR.ORG

Intervention

Control

Simple Randomization of

Districts

10 Million Death Study (10MDS) Rapid Mortality Data from 60 developing countries

• Build on successful Indian MDS • Open source data (edX course: Death 101, May 2014) • Low cost: <$1/household, sustainable as work with Censuses

Today 2020

Jha, BMC Med 2014

Conclusions

1. Counting the dead and describing causes are central to reducing premature mortality in urban and rural India in the 21st century

2. Three low-cost, high-impact applications of science relevant to urban and rural India:

- Child survival - Malaria control - Tobacco control