1 UNDERSTANDING
DATA IN THE
WORLD HEALTH
STATISTICS SERIES
Since 2016 the World Health Statistics series has served
as WHO’s annual report on the health-related Sustainable
Development Goals (SDGs). The effective monitoring of
SDG indicators requires comprehensive national health
information strategies based on the use of data from
sources such as civil registration and vital statistics systems,
household and other population-based surveys, routine
health-facility reporting systems and health-facility surveys,
administrative data systems and surveillance systems. Some
indicators also rely on non-health-sector data sources.
Making sense of the often complex available data on health
indicators can be highly challenging. Health data derived
from health information systems, including health-facility
records, surveys or vital statistics, may not be representative
of the entire population of a country and in some cases may
not even be accurate. Comparisons between populations
or over time can also be complicated by differences in data
definitions and/or measurement methods. Although some
countries may have multiple sources of data for the same
year, it is more usual for data not to be available for every
population or year. For example, measurement frequency
for data collected through household surveys is typically
every 3–5 years. This means that the years for which data
are available differ by country. To overcome these and
other issues and allow for comparisons to be made across
countries and over time, analysts develop mathematical
and statistical models with the aim of producing unbiased
estimates that are representative and comparable.
In World health statistics 2017,1 more than 50 health-
related SDG indicators were identified. Currently, sufficient
monitoring data are available for 36 indicators and these
data are presented in Annexes A and B of the current report,
as well as online in the WHO Global Health Observatory
(www.who.int/gho/en). For most indicators, comparable
estimates are reported if they are available. Such data have
been generated using a database of primary data and a
mathematical or statistical model, followed by consultation
with the relevant WHO Member State. In these cases,
the database of primary data used to derive the estimates
is available online, together with other documentation
required by the Guidelines for Accurate and Transparent
Health Estimates Reporting (GATHER).2
For other indicators, the most recent observation from
a database of primary data is reported. Primary data is
1
World Health Statistics 2017. Geneva: World Health Organization; 2017 (http://www.who.
int/gho/publications/world_health_statistics/2017/en/, accessed 28 March 2018).
2
Stevens GA, Alkema L, Black RE, Boerma JT, Collins GS, Ezzati M et al. Guidelines
for Accurate and Transparent Health Estimates Reporting: the GATHER statement.
Lancet. 2016;388(10062):1–5 (https://www.researchgate.net/publication/304576854_
Guidelines_for_Accurate_and_Transparent_Health_Estimates_Reporting_The_GATHER_
statement, accessed 28 March 2018).
MONITORING HEALTH FOR THE SDGs 1
 2
Fig. 1.1
Schematic overview of WHO data compilation, processing, analysis, consultation and reporting

DATA COLLECTION DATA COMPILATION COMPARABLE CONSULTATION REPORTING PUBLICATION AND

1 2 AND VALIDATION 3 ESTIMATES 4 5 6 DISSEMINATION
WHO collects data from a While some primary data need A statistical or mathematical After initial statistics are Data and methods are WHO publishes its official
variety of sources through to be compiled, processed and model is used to calculate obtained, Member States reported in line with statistics through its flagship
official requests or accessing validated, other primary data comparable estimates. are given an opportunity to GATHER. products such as the World
publicly available data. are forwarded for publication. comment on methods or Health Statistics series.
provide new primary data.

WORLD HEALTH STATISTICS: 2018

Some primary data are forwarded for official publication

A country consultation may be carried out

Via official request

Publicly available data

ARABLE ESTIMAT
MP ES
CO
CONSULTATION

Reporting in line with GATHER

Official publication
of health statistics

Key informant data may be validated before official publication

Note: Data on emergencies does not pass through the country consultation and reporting process in most cases
Table 1.1
Categories of data series appearing in World Health Statistics 2018

Label Definition Comparability
Comparable estimates A statistical or
mathematical model
was used to generate
comparable statistics for
each country on the basis
of available primary data.
Primary data A compilation of summary
statistics based on
empirical measurements,
for example statistics
from individual surveys
or case notification data.
These may include raw or
processed data.
Other data Data which are neither
primary data nor
comparable estimates
(usually key informant
data).
Evidence base Member State consultation Example (SDG indicator)
Statistics mean the same Comparable estimates are Member States are Maternal mortality ratio
thing in different countries. reported for countries with provided with draft (3.1.1)
primary data, as well as estimates, and may
for countries with weak or provide comments on the
no primary data. methods and data used.
Country data are typically If statistics are reported for Although Member State Prevalence of stunting
from different years, and a country, they correspond consultation is not among children under 5
data years may differ to primary (empirical) required, some data series years of age (2.2.1)
by up to 10 years. Some measurements from the are consulted upon with
data series include only last 10 years. Member States.
statistics which are
collected using the same
measurement methods
and calculated using the
same indicator definition,
while other data series
include statistics collected
and calculated in a variety
of non-comparable ways
(non-comparable statistics
are identified by footnotes
in the annexes).
Statistics may not mean Statistics are reported Member State consultation Average of 13 International
the same thing in different regardless of primary data is not required; these data Health Regulations core
countries. availability. are usually provided by capacity scores (3.d.1)
Member States.

an umbrella term that includes both raw data (measures
derived from primary data collection with no adjustments
or corrections) and processed data (calculated from raw
data).1 Processing raw health data can include cleaning data
by removing implausible values, calculating an indicator
with an algorithm or adjusting a statistic for bias. In some,
but not all, cases these data have been consulted upon with
each respective Member State.
Although most data series reported in World Health
Statistics are either compilations of primary data or
comparable estimates, there are some data series which
do not clearly fit into either of these categories. Typically
these are data series compiled using the results of surveys
of key informants, such as government officials, in countries.
Such data series may reflect primary data known to the
informant, estimates known to the informant, or the opinion
of the informant regarding the local situation. In order to
label such data in the current report, a third data category
¬ other data ¬ is used.
A schematic overview of the compilation and processing
of primary data, calculation of comparable estimates,
consultation with Member States and publication in
the World Health Statistics and other World Health
Organization data products is provided in Fig. 1.1.
In World health statistics 2018, each data series has for the
first time been labelled as “comparable estimates”, most
recent “primary data” or “other data” to clearly indicate the
category to which it belongs. The features of each of these
three types of data series are outlined in Table 1.1. These
data labels can be used by readers of this report to guide
interpretation of the data presented and to inform further
investigation on data sources by topic. Users of comparable
estimates should interrogate the availability and quality of
the underlying data used to generate the estimates, and
should take into account uncertainty intervals (available
online at the WHO Global Health Observatory). Users
of primary data should assess whether the data are
comparable, taking into account the inclusion/exclusion
criteria for the database, whether adjustments were made
to improve comparability and the year of data collection.
In this regard, attention should be given to the footnotes
on country statistics provided in Annex B. Finally, users of
statistics which are labelled as other data should be aware
that primary data may not be available, and that data are
often not comparable across countries.
In addition to the importance of understanding these
different types of information at the global level to inform
interpretation and policy dialogue, the reviewing of data
sources and data availability at country level can also help
to define the scope of ongoing and future health information
strategies. In particular, any gaps in data collection can
be identified and solutions prioritized to support the
development of informed national health strategic plans.

Stevens GA, Alkema L, Black RE, Boerma JT, Collins GS, Ezzati M et al. Guidelines
1

for Accurate and Transparent Health Estimates Reporting: the GATHER statement.
Lancet. 2016;388(10062):1–5 (https://www.researchgate.net/publication/304576854_
Guidelines_for_Accurate_and_Transparent_Health_Estimates_Reporting_The_GATHER_
statement, accessed 28 March 2018).

MONITORING HEALTH FOR THE SDGs 3