SAMPLE
REGISTRATION
SYSTEM
IN INDIA
Experience of Verbal
Autopsy
1
Sample Registration
System (SRS)
SRS initiated by the Office of the
Registrar General, India in 1964-65 on
a pilot basis and became operational
on full scale from 1969-70.
One of the largest continuous
demographic household sample
survey in the world covering 1.2 million
households and 6.3 million population.
2
SRS is a dual reporting system with
continuous and retrospective recording of
events by two independent functionaries.
The main objective of SRS is to provide
reliable annual estimates of birth and death
rates at the State and National level
separately for rural and urban areas.
SRS also provides data for estimating
Infant Mortality Rate (IMR), Total Fertility
Rate (TFR), Maternal Mortality Rate and
other measures of fertility and mortality.
3
Sample design
The sample design adopted for SRS is a uni-
stage stratified simple random sample without
replacement.
Stratification in rural area: In rural area, each
district within a State has been divided into two
strata viz. strata-1: villages with population less
than or equal 1500 and strata-2: villages with
more than 1500 population.
Stratification in urban area: In urban areas
stratification has been done on the basis of the
size of towns/cities. The towns/cities are
grouped into six size classes.
4
A simple random sample of
enumeration block is selected without
replacement from each of the size
classes of towns/cities in each State/Ut.
The sample unit in rural areas is a
village or a segmented village whereas
in urban area, it is a census
enumeration block.
5
Estimation Procedure
The estimates of population, live births,
deaths and infant deaths are obtained
using unbiased method of estimation.
The annual estimates of births, deaths
and infant mortality rates are based on
about 1,50,000, 50,000 and 10,000
reported number of sample births,
deaths and infant deaths respectively at
the national level.
6
INFANT MORTALITY
• Infant mortality is the most sensitive index of the level
of socio-economic development and the quality of life. It
is commonly used for monitoring and evaluating
population and health programmes and policies.
• Infants (less than one year) and early childhood (less
than five years) deaths still form a large fraction of the
total deaths (all ages).
• In India one out of every fifth death is of infant and a
total of about 1.8 million infants are dying annually
(based on IMR of 2002) as compared to 2.6 million in
1971.
• The proportion of infant deaths among early childhood
deaths is much higher and is over 70 per cent.
7
Decadal Trend of IMR in India
Significant decline in IMR during the last
three decades. The present level of IMR is
about one-half as compared to 1971.
The decades of 1970‟s and 1990‟s have
witnessed a decline of more or less of similar
order (10-11 per cent). The decline was
gradual during 1991-2000 as compared to
1971-80.
During 1981-90, the decline in IMR was
steeper, compared to preceding and
succeeding decade, and was about 17 per
cent.
8
COMPARATIVE DECLINE IN DECADAL IMR
(Based on three years moving average)
105
100
IMR Index
1971-80
95
1981-90
90 1991-00
85
80
1 2 3 4 5 6 7 8
Year
9
State - Scenario
The decadal IMR vary considerably from one
State to another ranging from Kerala(51) to
UP(176) during 1971-80, Kerala(28) to
UP(135) during 1981-90, and Kerala(15) to
Orissa(104) during 1991-2000.
The lowest levels of IMR have been recorded
by Kerala, Karnataka and Maharashtra during
the decade of 70‟s and by Kerala,
Maharashtra and Punjab in that order during
the last two decades.
The highest levels of IMR have been retained
by UP, MP and Orissa with some changes in
inter-se positions.
10
Sub-State level Variations
The existing sample size of SRS does not
allow small area estimation of IMR or
mortality analysis by socio-economic
status. IMR varies widely from one-region
to another. Thus, reduction in average
IMR in a State does not provide a
complete picture of mortality decline,
necessitating identification of high
mortality prone areas and planning
innovative strategy for its reduction.
11
IMR estimates - regional level, M.P.
(Based on three years moving average)
140
120
100 1995-97
80 1996-98
60
1997-99
40
20
0
Chhatisgarh
Northern
Central
Central
Pradesh
Plateau
Vindhya
Western
Madhya
Malwa
South
South
12
13
o o o o o o o o
68 72 76 80 84 88 92 96
AFGHANISTAN
o
o 36
36
INDIA
NATURAL DIVISION WISE
INFANT MORTALITY RATE (RU RAL) 1997-99
N
Boundary, International . . . Boundary, State/U.T . . .
A
Boundary, Natural Divisions . . .
o o
T
32 KILOMET RES 32
C 100 50 0 100 200 300 400 500
S
H
I
T
I A
K
N
I
N
B T o
o E 28
A
28 E
P
P
A BHUTAN
L
o
o BAN GLADE SH 24
24
MY ANM AR
I N D I A (BURM A)
>
--
o
--
New Moore I .
o
20
20 (I NDI A )
ARA BIAN
SE A BAY
OF
BENGAL
o o
16 16
Coc o I s .
L
(B URMA)
A
Narc ondam I .
K
A N
(I NDI A )
S
INFAN T M O RTALITY RATE
D A
H
BY N ATURAL DIV IS IO N
A
o
MA
o Barren I.
12
D
12 (I NDI A )
L
50.0 AN D BELO W
W
N
A
E
50.1 - 60.0
E
K
A N D A M A N S E A
P
S
60.1 - 70.0
A
S
H
(I
I N
N
E
D
A
D
70.1 - 80.0
(I
A
I A
N I
D
N
)
)
D
C
W
80.1 AN D ABO VE
I A
O
)
E
B
DAT A N O T A VA IL AB LE
A
E
R
o
P
o 8
8
I S
SRI
L A
N
LANKA
D
S
INDIRA P OINT
I N D I A N O C E A N
o o
72 EAST OF GREENWICH 76 80 o 84 o 88 o 92 o
Based upon Survey of India map with permission of the Surveyor General of India.
The territorial water s of India extend into the sea to a distance of
twelve nautical miles measured fr om the appr opr iate base line.
C Government of India copyright, 2003.
14
IMR - Components
The categorisation of IMR into neo-natal and Post
neo-natal rates helps in refining the strategies for
combating the infant mortality.
During the last decade (1991-99), the average
contribution of the neo-natal and post neo-natal
mortality to IMR has been recorded about two-third
and one-third respectively.
Over the three decades the decline in neo-natal
and post-natal mortality is 37 per cent and 60 per
cent.
Higher share of neo-natal mortality in IMR coupled
with lower decline and higher percentage of non-
institutional deliveries (about 75 per cent) suggest
strategies targeting especially the neo-natal deaths
and providing medical facilities for non-institutional
deliveries and home care of neo-nates. 15
Proportion of IMR, NMR and PNMR
(Based on average rate during 1991-99)
Infant
Post-natal
mortality rate
mortality rate
50%
18%
Neo-natal
mortality rate
32%
16
Components of neo-natal mortality
The categorisation of neo-natal mortality
(<29 days) into early (<7 days) and late
(7-28 days) neo-natal mortality
facilitates in refining the strategies for
reducing IMR.
Early neo-natal deaths constitute a
major chunk of deaths of neo-nates and
infants, and is as high as three-fourth of
neo-natal deaths and one-half of all
infant deaths.
17
Trends in early and late neo-natal
mortality during the 90's
105
100
95
90
Index rate
85
Early
80 neo-
75 natal
70 Late
neo-
65
natal
60
1990-92
1991-93
1992-94
1993-95
1994-96
1995-97
1996-98
1997-99
18
The decline in late neo-natal is 5 times
more than early neo-natal during the
last decade.
The early neo-natal has remained more
or less stagnant depicting a low annual
decline of about 0.5 per cent.
Interventional strategies appears to
have very little impact on early neo-
natal as compare to late neo-natal.
19
Maternal Mortality
Data on Maternal Mortality is available :
National Family Health Survey (NFHS)
• NFHS-1 (1992-93) : 424
• NFHS-2 (1998-99): 540
(maternal deaths per 100,000 live births)
Sample Registration System (SRS)
• SRS (1997) : 408
• SRS (1998) : 407
Estimates have large sampling fluctuations due to
inadequate sample size (based on about 600 maternal
deaths in SRS).
The available estimates indicates that about 100,000
women in India die every year from causes related to
pregnancy and child birth.
The finding suggests : anti-natal care for all pregnant women
and deliveries take place under hygienic conditions.
20
Revision of SRS sampling frame-2004
10-yearly based on recent results of Census
• To make necessary modification in the sampling
design
• To give wider representation of population
• To overcome the difficulties/limitations in the existing
scheme
• To meet the additional requirements
Enhancing the scope of SRS
Rationalisation of SRS forms
Better netting of events
User friendly
Easy for scanning
Streamlining the system
21
Features of the New SRS
Apart from reliable estimates at the state and national
levels for birth rate and death rate separately for rural
and urban areas, the new SRS will provide vital rates at
NSS Natural Division level (which is a group of
contiguous districts) for rural areas.
It will also provide reliable estimates of IMR at NSS
Natural Division level for rural areas.
Use of female literacy as a stratifying factor.
Separate estimates for four metros viz. Delhi, Kolkata,
Chennai & Mumbai.
Introduction of Verbal Autopsy instrument for
determining the cause specific mortality by sex and age.
The sample size of new SRS is enhanced from 6671 to
7597 units, covering about 1.4 million households and
over 7 million population.
22
Enhancing the scope of data
• Morbidity data
• Family planning practices data
• Data on abortion
• Personal habits – use of pan, tobacco, alcohol, food habits:
veg/non-veg
• Birth history of all currently married women in reproductive span
• Data on reasons of migration
• Data on school attendance (up to 16 years)
• Data on disability
Introduction of VA forms for recording structured
information and narrative for determining the cause
specific mortality by sex and age.
• Verbal Autopsy (VA) is an investigation of train of events,
circumstances, symptoms and signs of illness leading to death
through an interview of the relatives or associates of the
deceased.
23
Introduction of Unique Identification
Code
One of the significant modification
proposed is introduction of unique
identification code. This will result in :
easy storage and retrieval of data
aggregation at different levels
Cross-classification of various determinants
with fertility and mortality indicators
Cohort studies
24
Status of new SRS
The Baseline Survey for the new SRS is in
progress since Nov‟2003.
The urban Baseline Survey has been completed
in most of the states and the rural is in progress.
It is expected to complete the Baseline Survey
by March‟2004.
The effective date for the new SRS frame is 1st
January‟2004.
The first report based on new sample containing
vital rates for 2004 would be available in 2005.
25
TARGETS
The National Population Policy has set the targets of
reduction in IMR to 30 and MMR to 100 by 2010.
The goal is to achieve 53 per cent decline in IMR
from its present level of 64 in 2002 in next 8 years.
For MMR the target is to achieve 75 per cent decline
from its present level of around 400 by 2010. The
appropriate strategies to achieve the above goals
have been formulated.
To monitor the impact of these strategies in
reduction of IMR and MMR, there is need for an
appropriate evaluation system.
Whether the existing SRS will continue to be
appropriate to map the decline in IMR and MMR? If
not, then what more is expected from SRS.
Suggestions are welcome.
26
CAUSES
OF
DEATH
IN
SRS 27
Importance of Causes of Death Data
Data on causes of death are useful for
health planners, administrators, and
medical professionals:
To identify the public health importance of different
diseases.
To make a decision on allocation of resources for
controlling various diseases.
To evaluate trends in causes of mortality over time in
order to assess the impact of national health
programmes.
To analyse the socio-economic, demographic and life
style factors that are associated with the deaths due
to various diseases.
28
DATA ON CAUSES OF DEATH
The data on causes of death is available from the
medically certified deaths occurring in hospitals
whether public or private covered under the scheme
of „Medical Certification of Causes of Death‟. It has
its own limitations.
Different stages of implementation in different states and uts.
Selected areas-only urban
Selected hospitals
(Does not provide cause of death profile at state level
for all urban deaths)
The “Survey of Causes of Death (Rural) ” has been
integrated with SRS from 1999 to cover all deaths
occurring in a nationally representative sample both
in rural and urban areas.
29
Inadequacies in Causes of Death Data
Cause of death mainly the respondent
perception.
Instruments and procedures not well
developed.
The cause of death assigned by the SRS
Supervisor based on symptoms list.
No physician review was involved.
SRS Supervisors not fully trained.
30
Verbal Autopsy Activities
Part Time Enumerator
Continuous recording of birth/death events
Inform households about
the conduct of VA
Supervisors
Half yearly retrospective survey
Collection of the circumstances, symptoms Cause of Death Assignment
Cause of Death
by Health Professional
and signs of illness and Narrative in
Assignment by
VA forms
Health Professional
Quality Check
10% Re-sample in the field
Cause of Death Assignment
by independent Re-Sample Teams by Health Professional
31
Initiatives in SRS
To improve the data on causes of death in SRS the
following initiatives were taken in recent past:
Development of VA Forms
• Forms were developed based on the existing
experience of WHO, Chinese Surveillance System
and other international and national studies.
Type of Forms : incl. Structured & Narrative
• Neo-Natal Form
• Childhood Form
• Adult Form
• Maternal Death Form
32
Initiatives in SRS
Conduct of pre-tests of VA Forms in various
regions
Review of the results of pre-test by eminent
epidemiologist/researchers
Refinements in VA Forms based on the feed-
back
Preparation of VA Instruction Manuals
Standardized sandwich training to 800 RG
Supervisors on VA methods by leading
institutions
33
Initiatives in SRS
Premier Institutes like CGHR (University of
Toronto), NIMHANS (Bangalore), PGI
(Chandigarh), ICMR, TIFR (Mumbai), Medical
colleges of India, ERC (Chennai), have been
identified in all the major States as long term
technical partners with SRS for :
Training/Refresher Training to RGI Staff on verbal
Autopsy
Conducting VA in 10 percent resample units
Assignment of causes of deaths (double
assignment)
Quality Control
Epidemiological analyses
34
Initiatives in SRS
Re-sampling
Objective- 10% of VAs for each SRS
Supervisor will be checked by
collaborating institutions for training
feedback
Identification of operational problems
and possible remedial measures:
Physician coding
100% double coding,
Reconciliation with another physician
Adjudication of disagreements
35
Preliminary results of VA on Causes of Death
Causes of death (using WHO groupings) Male Female Total
I. Communicable diseases, maternal and
perinatal conditions and nutritional 32 41 36
deficiencies
II. Non- communicable conditions 39 31 36
III. Injuries 10 5 8
IV. ILL-defined 19 22 20
All causes 100 100 100
36
Preliminary results of VA on Causes of Death -
Communicable diseases, maternal and peri-natal
conditions and nutritional deficiencies
(in numbers)
Male Female Total
Causes of death (using WHO groupings)
Tuberculosis 72 55 127
Other infectious diseases 63 67 130
HIV 4 0 4
Diarrhoeal diseases 116 187 303
Childhood-cluster diseases 34 37 71
Respiratory infections 81 101 182
Maternal conditions - 52 52
Peri-natal conditions 170 112 282
Nutritional deficiencies 19 32 51
Total 559 643 1202
37
Preliminary results of VA on Causes of Death –
Non-communicable conditions
WHO Grouping Male Female Total
Cardiovascular 244 152 396
All Cancers 74 57 131
Other non- 227 154 381
communicable
Respiratory diseases 177 124 301
Total 722 487 1209
38
Preliminary results of VA on Causes of Death –
Injuries
WHO Grouping Male Female Total
Unintentional injuries 146 59 205
Self-inflicted injuries 12 19 31
Other intentional 28 3 31
injuries
Total 186 81 267
39
Present Scenario & Future Plans
All the SRS Surveyors have been trained
and re-trained in the art of canvassing VA
The VA has been introduced in all the
states/uts. as an integral component of
SRS
The preliminary results for two Half Yearly
Surveys (2nd HYS, 2002 & 1st HYS, 2003)
were presented in Trivandrum Workshop
The results suggest that VA would result in
generating cause specific mortality by age,
sex and other risk factors on a continuous
basis. 40
41