1. Sources of Vital statistics Data
Data analysis and Report writing workshop

2. What are vital statistics?

3. Population and institution based sources
Population-based data sources are those that are representative of the whole population.
population surveys,
censuses
civil registration.
Institution-based data sources are those collected routinely from administrative and operational activities,
Health Information Systems (HIS)
hospital discharge data,
police records for attended deaths,
social security records.
health facility surveys, where data are collected in an institution.

4. Routine vs period collections
Routinely collected administrative data
CRVS
HIS/ Hospital information
Health facility data
Periodic collection
Censuses
Surveys

5. Census collections Primary source of mortality data in PICTs
 A population census is a compulsory, universal and simultaneous enumeration of the national population, conducted on a periodic basis.
Provide population denominator data AND mortality data
 Information on births and deaths may be derived
directly
indirectly
from changes in population (by age and sex) based on survival and migration
reliant on accurate migration data

6. Census
...a compulsory, universal and simultaneous enumeration of the national population, conducted on a periodic basis
Censuses provide denominators (population) for computing fertility and mortality indicators
Can also provide numerators (births and deaths), and data for indirect estimation of mortality through a question on retrospective events

7. Calculating mortality from a census
Direct measures
ask about births and deaths in the household over a nominated reference period –
usually within the last 12 months.
Take a complete birth history (not usually used in the census)
 Indirect measures
collect some information on births and deaths then approximate when these events would have occurred.
Children ever born –children surviving
Orphanhood
Widowhood
These measures can then be used as inputs for model life tables that generate estimates of age-specific measures of mortality and LE.
Models that use a single input parameter based on childhood mortality have been shown to have a tendency to under-estimate mortality in PICTs

8. Periodic Surveys
i.e. Demographic and Health Surveys (DHS) and the UNICEF Multi- Indicator Cluster surveys (MICS).
not ongoing data collections, but are collected in a specific period of time.
collect data for a proportion of the population considered to be representative of the broader population of interest.
 may collect data to undertake direct estimation of mortality from reported deaths, or indirect estimation based on the age distribution methods described previously.
The reliability of estimates from survey data is driven by how well the sample selection reflects the broader population of interest
Surveys are also subject to
recall bias, and
response bias
Repeated household surveys may also be used to estimate mortality based on survival and migration

9. Birth history questionnaire
Typical birth history, from the Demographic and Health Survey conducted in 1999 in zimbabwe. Most major survey initiatives will collect a complete birth history; the DHS in particular is a major source for direct estimations from birth histories. Usually will allow up to 20 entries, depending on the country context. Using this data, you can calculate the risk of dying before age 1 or age 5 (IMR or child mortality). Data crunching part can be tricky, but the concept is remarkably straightforward:
Count the number of children that died before age 1 or age 5.
Count the total number of children that were born.
Divide, and presto: IMR or CMR.

10. Problems with using birth histories
Recall
Reference period
Inaccurate reporting from respondents
Sample size in low mortality countries
Cultural appropriateness
Length of the questionnaire
Recall: Mothers may not recall every birth correctly; birth histories seem to improve recall, but there may still be problems with women, for example, reporting a stillbirth as a death (which they should not) or not reporting a child who died very shortly after being born. Some of these concerns are problems in all sources of data (i.e. stillbirth vs. birth followed quickly by death)
Reference period: because birth histories are for the entire reproductive span of the woman being asked, the mortality rates gleaned from the analysis are, in fact, average rates over a particular period.
Censoring. Calculating under 1, or under 5 deaths requires you to “censor” your data: for example, if you are examining under-5 mortality, you can’t include any children in your analysis who were born less than 5 years ago, as you don’t know yet if they will have made it to their 5th birthday. This means that your estimate is older than you might like. For this reason, analysts will sometimes calculate under-2 and under-3 mortality instead of under-5, as it yields more recent results but is still a close approximation of under-5 as most childhood deaths appear to occur in the first 3 years of life.
Sample size, in low mortality countries, can mean that you don’t have enough children dying to make an accurate estimate.
Cultural appropriateness: both in terms of asking unwed women about their reproductive history (samples often need to be limited to married women of reproductive age, thus not truly representative), and asking about dead children (i.e. hill tribes in thailand will not speak of deaths).
The major reason that birth histories are not included into a survey or census instrument is simply that there are too many questions and it will take too long to ask women about all of their births and related details. In such a situation, it is still possible to estimate child mortality, it just requires n

11. Routine vital registration collections
Civil registration provides a legal basis for the recording of vital events such as live births & deaths
In most developed countries, it is a legal requirement that:
a medical practitioner completes a death certificate whenever anybody dies
vital events, such as births and deaths, are registered
An efficient routine CRVS system, with medical certification of CoD, provides ongoing and relatively low cost data collection and therefore timely mortality data for decision making
upheld as the “gold standard” for birth and death data

12. medical certificate
considered the reference standard for CoD information, in the absence of an autopsy, as a qualified practitioner is required to assess the case and make an informed decision concerning the sequence of events that led to the death

13. Civil Registration
Dual function – establish civil status, collect data for planning
Citizen’s / government responsibility
Passive system, as opposed to special registration activities
Multiple government agencies responsible
Data transmission losses, need for inter-sectoral collaboration
Difficulties in cause attribution in some situations
Data may not be timely without strong incentives to report
As the legal record, civil registration of vital events is frequently legislated as the official source for mortality data despite not always being the most reliable data source in country.
Discuss incentives

14. Core functional areas for CRVS and underlying support needs
For good evidence for policy and planning – all of the components must be functioning well
Strong CRVS systems are multi-sectoral

15. Figure 5.1: Diagram of the reporting and registration processes for deaths
Local Government Office (Island Council etc)
Civil Registrars Office (Ministry of Justice)
National Statistics Office
Ministry of Health
Family
National Planning Office / Process

16. Example Routine Death Reporting System
Town Officer
Civil Registrars Office (Ministry of Justice)
Death Certificate
National Statistics Office
Family
National Planning Office / Process
Health System
Doctor* Nurse#
Health Public
Information Health
System Nursing
Medical Certificate of Death
Monthly report
Notice of Death
District Officer
Prime Ministers’ Office
Quarterly report

17. Health registration of vital events
 Vital events such as births and deaths are also often recorded through routine data collections within the health system.
health data collections are primarily to inform operational decisions, and CoD is central to this purpose.
may include
medical certificates
community nursing reports,
facility based data
Must consider if this is population based data or facility based data

18. Proportion of estimated deaths reported by Island group Tonga : Reconciled data – unadjusted

19. Proportion of estimated deaths reported by gender, source and period, Tonga : Two-source analysis

20. Demographic surveillance sites
A demographic surveillance system captures all vital events in a specified area.
Often combined with disease detection (sentinel surveillance)
used in PNG
unlikely to be a suitable solution in other PICTs with small populations

21. Data collection (mortality level data)
Summary of Population based Data Collection Approaches for Mortality Data
Data Source
Periodicity
Sample frame
Period of interest
Data collection (mortality level data)
CoD Data Collected?
Census
Periodic –
5-10 years
Whole Population
Retrospective
Direct - (deaths in the household)
Indirect – partial birth history (CEB/CS) & orphanhood data No
Survey
DHS
~ 5 years
Selected sample – representative of whole population
Direct – complete birth history
MICS
~ 5-10 years
2 stage clustered sample – representative of whole population
Indirect - partial birth history (CEB/CS)
Other household based surveys
Usually once-off
Varies
Possible - using verbal autopsy
Routine vital registration
Civil Registration
Continuous
Whole population (depending on coverage)
Current
Direct reporting of event
Yes
Health vital registration
Hospital discharge records
Hospital cases only
Other routine databases
Various
Varies – usually targets sub-population of specific interest.
Usually limited
Demographic surveillance Sites
Selected areas – usually not representative of whole population over time.
(source: Carter, 2013)

22. Institution-based data sources
 Institution-based data sources in the health sector primarily include those based at hospitals and health centres
 In the hospital setting, unit record data on deaths and CoD may be collected through hospital separation data
(hospital records that indicate whether the patient was discharged home from hospital, transferred to another facility or died).
CoD is based on the principal diagnosis at the time of death
Countries may also collect data on deaths through primary or community health care nursing programs,
deaths recorded on a separate form to be sent to the local area nursing manager and/or
recorded on monthly reports.

23. Registries source of outcome data
i.e. – cancer registries, pregnancy registers
May be population or institution based
Population registries much more common where there is national testing facilities
tend to be more useful for measures of morbidity rather than mortality (as they are generally a secondary data source in this case).
For deaths: registries record anyone who dies WITH the disease rather than FROM the disease (as the designated underlying cause)
therefore the data is not directly comparable to that obtained from vital registration.
 Pregnancy registers often maintained at health facility level, but rarely electronic and generally not centrally collated.

24. Data sources to ascertain causes of death
pathological autopsy report (gold standard)
physician certification at time of death
physician certification in absentia, using medical records
lay assignment of cause
verbal autopsy – interview with relatives of deceased to identify symptoms leading to death 

25. Where does your data come from?
Think about the data you brought to the course and where it comes from…
Why was it collected?
How was it collected and by whom?
What are the strengths and weaknesses of that data collection?
What impact is that likely to have on your data?