Chapter 12. Population-based data sources

Overview

Population-based data sources, such as household surveys, civil registration and vital statistics (CRVS) systems and censuses, are intended to be representative or completely inclusive of a defined population. They are important sources of data for health inequality monitoring. They contain information either from a representative sample of the population or from every individual in the population.

Household surveys are carried out in probabilistically selected samples of the population, meaning everyone in the population has a given chance of being selected. Household surveys are conducted at a particular point in time (yielding cross-sectional data) and may be repeated at regular intervals.

CRVS systems and censuses are designed to gather data systematically from every member of a population – although in many countries, these sources are not fully functional. CRVS systems collect data on an ongoing or rolling basis. Censuses are undertaken periodically, according to a set schedule (often every 10 years).

The aims of this chapter are to describe the characteristics of household health surveys, CRVS systems and censuses, and discuss how each can be useful for health inequality monitoring. The chapter addresses the strengths, limitations and key considerations for using these sources, and outlines possibilities for improvements to enhance their usability for monitoring inequalities. A better understanding of these data sources will prepare readers to assess their suitability for different applications of health inequality monitoring.

Household surveys

Household surveys are rich sources of disaggregated data and are well suited for inequality analyses (1). Some household surveys are general in nature and may contain little or no health data. Other household surveys focus specifically on health or health determinants. For the purposes of health inequality monitoring, household health surveys – which collect detailed information centred around one or more health topics – are of particular importance, although other types of household survey (e.g. labour force surveys, income and living condition surveys, household budget surveys) may also contain relevant information.

Regardless of their topical focus, household surveys share several general characteristics. They collect data from a sample of individuals or households within the population rather than every individual or household within the population. This is termed a probabilistically selected sample when the likelihood of being selected is known. In most cases, surveys rely on complex survey sampling designs, which involve selecting the sampling units via multiple stages or phases.

When a sample is selected appropriately, it can provide information that is statistically representative of the entire population from which it was drawn. National household surveys are designed to be representative at a national or subnational level. Household surveys are not, however, typically designed with the purpose of having sufficient sample sizes in all population subgroups of interest for health inequality monitoring. To account for this, some surveys may oversample one or more subgroups of the general population that would otherwise be too small for disaggregation and analysis. Sampling design characteristics, including stratification, clustering, multistage sampling and weighting, need to be taken into consideration when analysing data from surveys (see Chapter 17).

Household health surveys

Household health surveys usually cover a large number of health (or health determinant) indicators within the same survey, all related to a similar theme, such as reproductive, maternal and child health; nutrition; noncommunicable disease risk factors; or communicable diseases. They also contain questions related to background demographic, social and economic characteristics of the respondents. Household health surveys may cover multiple themes across different modules (although some modules may not be included in every round of the survey).

Household health surveys typically gather self-reported information through interviews or self-administered questionnaires. Surveys might focus on understanding health status, use of health care, and health-related behaviours. They may also collect more objective information about health metrics through physical examinations and biomarkers (e.g. for assessing diabetes, HIV infection or anaemia). Although household health surveys are conducted in most countries, they tend to be a particularly important data source for health inequality monitoring in low- and middle-income countries, where data from CRVS or other sources may be less available or less reliable (see Chapter 5). Many countries conduct their own national household health surveys, but other surveys are administered by donor organizations, nongovernmental organizations or international organizations.

Multicountry household health survey programmes apply consistent methodologies to collect comparable data across multiple countries. Data from multicountry household health surveys can be used for benchmarking – that is, for comparing within-country inequalities between different settings and populations to get a broader understanding of the state of inequality. Two well-established multicountry household health survey programmes are the Demographic and Health Survey (DHS) Program of the United States Agency for International Development, and the United Nations Children’s Fund (UNICEF) Multiple Indicator Cluster Survey (MICS). Box 12.1 provides details about these and other multicountry household surveys across selected health topics.

BOX 12.1. Examples of multicountry household surveys across selected health topics

The following examples of prominent multicountry household health surveys span different topics and settings. This list is not exhaustive. More information about each survey programme is available online.

Adult health and ageing:

• The WHO Study on Global AGEing and Adult Health (SAGE) is a longitudinal study focused on adults aged 50 years or over (2). The study collects data from nationally representative samples in six countries. It contains information about the health and well-being of adult populations and the ageing process.

• The Survey of Health, Ageing and Retirement in Europe (SHARE) studies the effects of health, social, economic and environmental policies over the life course of citizens across 28 countries in Europe (3).

Malaria:

• The Malaria Indicator Survey (MIS) is a standalone household survey that collects data on bed-net ownership and use, prevention of malaria during pregnancy, and prompt and effective treatment of fever in young children. Data collection may also include biomarker tests for malaria and anaemia. It has been conducted in over 30 countries (4).

Noncommunicable diseases:

• The WHO STEPwise Approach to NCD Risk Factor Surveillance (STEPS) surveys cover key behavioural risk factors (tobacco use, alcohol use, physical inactivity, unhealthy diet) and biological risk factors (overweight and obesity, elevated blood pressure, elevated blood glucose, abnormal blood lipids), with expanded modules on other topics such as oral health, sexual health and road safety (5). Since 2002, over 130 rounds of the survey have been conducted across all world regions and country income groups (6).

Reproductive health:

• Reproductive Health Surveys (RHS) provide information about various aspects of reproductive health, including antenatal care, fertility, contraceptive awareness, knowledge and use, and sexually transmitted infections (7). RHS were first conducted in 1975, when they were known as Contraceptive Prevalence Surveys. In more recent years, data collection has focused on eastern Europe and Latin America, although data are available for countries in other regions (8).

• Performance Monitoring for Action (PMA) surveys include key indicators of family planning use, water access, sanitation and health. They cover 11 countries (9).

Reproductive, maternal, newborn and child health:

• DHS have been conducted in over 90 countries, covering diverse topics and dimensions of inequality relevant to reproductive, maternal, newborn and child health. Standard DHS are typically conducted about every five years (10).

• The MICS Program spans 120 countries, providing internationally comparable data on women and children (11).

Tobacco:

• The WHO Global Adult Tobacco Survey (GATS) is a nationally representative household survey that collects information on prevalence of tobacco use, secondhand tobacco smoke exposure and policies, cessation, knowledge, attitudes and perceptions, exposure to media and economics (12). It has been implemented in more than 30 countries (6).

• The WHO Global Youth Tobacco Survey (GYTS) collects information from students aged 13–15 years to monitor tobacco use among youth, and to guide the implementation and evaluation of tobacco prevention and control programmes (13). More than 500 rounds of the GYTS have been conducted in over 180 countries, territories and areas (6).

Strengths and limitations of household health surveys

A general strength of household health surveys for inequality monitoring is the inclusion of detailed information on an array of health (or health determinant) indicators and dimensions of inequality at the individual or household level, within the same dataset. This presents opportunities for disaggregation by diverse inequality dimensions. The use of common and well-documented methods to measure health indicators and inequality dimensions (including indicator frameworks and criteria) makes the estimates more reliable. It also helps to ensure the estimates are consistent and comparable across settings. Because data are collected at the individual level, they may be well suited for complex inequality analyses, such as multiple regression and compound vulnerability and advantage assessments (see Chapter 25).

Household health surveys are also advantageous in terms of their versatility to adapt to changing data needs. Many surveys are scheduled to be conducted on a recurring basis, such as every three to five years, but they may be done, in whole or part, on an ad hoc basis. Further, survey questions and methodologies can be updated and adapted between survey rounds to reflect emerging health issues, and selected survey modules may be included or excluded.

There are possible downsides to this versatility, namely that changing survey questions may limit the ability to compare like measures over time. Nevertheless, repeating surveys can generate comparable data that are useful for tracking changes in health inequalities over time.

Using surveys, it is possible to collect data pertaining to universally relevant metrics (which can be harmonized and compared across countries) along with integrating setting-specific considerations that reflect more local priorities and realities. For example, the DHS makes available several optional modules that may be added to complement the core content of the surveys. In 2022, three new optional modules included child well-being and household structure, human papillomavirus (HPV) vaccination and mental health (14).

Like all data sources, the quality of the component data is contingent on the design and implementation of the underlying methods. There may be cases where data from household health surveys are incomplete or inadequate for the intended inequality monitoring (Box 12.2). This may be related to the survey instrument design – for example, because certain dimensions of inequality were not captured in data collection instruments (e.g. measures of sexual orientation or gender identity) or certain subcategories were not reflected in the response options (e.g. ethnic minorities or specific religious affiliations). It may also reflect logistical challenges, such as certain regions being excluded because they are inaccessible due to conflict or other humanitarian crises. Response bias may be a concern, because people experiencing disadvantage tend to be undersampled – and although there are strategies to mitigate response bias, they may not allow for subsequent disaggregation.

BOX 12.2. Sampling and sample size limitations

A potential limitation to the use of household surveys in health inequality monitoring relates to the issue of small sample sizes for some population subgroups, depending on the dimension of inequality of interest. Household health surveys are generally designed to draw precise conclusions about the overall population, but they may not be representative of smaller population subgroups or remote or otherwise difficult-to-access geographical areas. This could be due to members of the small subgroups having a lower chance of being selected than everybody else. Additionally, subgroup estimates based on small sample sizes have high levels of uncertainty and are subject to suppression rules when displaying results.

Therefore, some surveys may not be suited for double or multiple disaggregation – which involves filtering data according to two or more dimensions of inequality simultaneously – because of sample size and sampling design considerations. These smaller subgroups may, however, be highly relevant for inequality monitoring. For example, the sample size of a nationally representative survey may be sufficient to estimate health indicators by subnational region (i.e. single disaggregation), although the sample size may not be sufficient to disaggregate further by a second dimension within those subnational regions, such as age, economic status or sex.

If it is anticipated that the survey will have smaller subgroups within a population, this problem can be mitigated by oversampling. This involves recruiting larger samples from smaller minority groups, even though these subgroups may represent a relatively small proportion of the overall population. Combining multiple years of data is another strategy that may be appropriate to address the problem of small sample size for a population of concern.

Other issues may arise if the data collection activities are poorly executed, such as interviewers not receiving sufficient training, or delays in data collection, processing and publishing (which may be more likely to occur during periods of uncertainty such as the COVID-19 pandemic). Depending on the timing of the inequality monitoring activity, the data may not be timely and therefore may be less useful for tracking the impact of ongoing equity-oriented interventions. On a practical note, household health surveys are resource-intense and typically expensive to carry out, and therefore they may be conducted less frequently or have limited sample sizes.

To enhance the completeness of the data source, additional improvements and investments in the data collection exercise may be required. To improve the relevance of the data and reduce differential response bias, survey modules should be developed in consultation with diverse stakeholders involved in and impacted by the collection and use of the data (e.g. conducting focus groups and piloting data collection instruments). Consultation and engagement with the people conducting inequality analyses may also be warranted (see Chapter 4).

Online repositories and tools have been developed to support the accessibility of household health survey data by enabling access to datasets, access to accompanying metadata (detailed information about how data are collected and how indicators are calculated), and, in some cases, access to data exploration and analysis features. The WHO Health Inequality Data Repository, for example, features several datasets sourced from multicountry household health surveys, including DHS, MICS and RHS (15).

Civil registration and vital statistics systems

CRVS systems are an essential part of a country’s administrative and statistical infrastructure, capturing crucial information about vital events within the population. They aim to collect vital event information about all members of the population on a continuous basis to contribute to, in combination with census data, key population statistics. These statistics serve as a cornerstone of public health and other population-level planning.

The scope of health information available from CRVS systems centres around births, deaths and causes of death. In countries where these systems are fully functioning, they register all births and deaths and compile other vital statistics, including cause of death information. Some CRVS systems also record marriages and divorces. They usually contain the minimum information about certain dimensions of inequality – age, place of residence and sex – and some also collect information such as occupation and ethnicity.

The operation of CRVS systems is designed to follow a series of steps with clearly demarcated roles and procedures (Box 12.3). As the name suggests, there are two components to CRVS systems – civil registration and vital statistics (17, 18). Civil registration is the legal component, whereby vital events are registered in accordance with the legal requirements of the country. In many cases, civil registration systems are intended to be continuous, permanent, compulsory and universal, adhering to strict national standards, providing the legal documents required by individuals during their entire life course. In other cases, however, registration may be required only for specific purposes, such as probate or obtaining certain benefits, and therefore is neither compulsory nor universal.

BOX 12.3. Operational components of CRVS systems

The Ten Milestones CRVS framework outlines 10 sequential steps required to ensure all births and deaths are reported, recorded, certified and incorporated into the vital statistics of a country (16). The civil registration subsystem begins with the notification process, which captures the minimum essential information related to births or deaths by a designated informant (Step 1).

Active notifications are sent directly to the local civil registrar for validation and official registration. Passive notifications require a family member to fill out a form and declare the event themselves (Step 2).

After validation and verification processes, the civil registration office formally registers the event (Step 3), and the information is stored in permanent archives (Step 4). Legal certificates are issued certifying the event (Step 5), and information is shared across government systems (Step 6).

The vital statistics subsystem entails aggregating and summarizing information on vital events, yielding a report of vital statistics. Data about vital events are compiled (Step 7), the quality is assessed (Step 8), and vital statistics are generated (Step 9). The statistics are published and disseminated – for example, as annual national statistics in a public repository accessible to users (Step 10).

Civil registration often falls under the ministry of the interior or justice or local government. The health sector, however, has a contributory role in strengthening CRVS systems, because health workers are usually present surrounding births and deaths and can support the timely and accurate reporting of related information.

Drawing from civil registrations and censuses, vital statistics systems compile and disseminate statistics pertaining to vital events of interest, such as live births, adoptions, legitimations, recognitions, deaths and fetal deaths, and marriages, divorces, separations and annulments (17, 18). As countries establish and build capacity for CRVS systems, they may progress from sentinel registration (at certain surveillance sites), to sample registration (capturing a representative sample of the population), to full registration (19).

CRVS data are important for public health decision-making, such as developing policies and planning services (20). They provide foundational information about fertility, mortality, life expectancy, burden of disease and emerging health needs. They may be a useful input to monitor inequalities related to these topics. For example, fertility statistics have implications for monitoring the need for family planning, school enrolment and immunization coverage, and for conducting epidemiological studies. In addition to their use in inequality monitoring, mortality statistics are used to understand health-care requirements, monitor interventions and prioritize health needs.

Strengths and limitations of data from CRVS systems

Once they are established, fully functional and covering all members of the population, CRVS systems serve as the most timely and reliable source of data on fertility, mortality and cause-of-death indicators. Data from CRVS systems, therefore, are particularly well suited for monitoring related to these notifiable events. Monitoring inequalities in life expectancy, for example, when the data are of adequate quality, can provide insights into the implications of underlying socioeconomic inequalities and risk factors associated with different causes of death among different subgroups (21).

For the purposes of health inequality monitoring between population subgroups, the health data from CRVS systems are powerful when linked with information on inequality dimensions. CRVS systems usually contain information on dimensions of inequality, such as age, ethnicity, Indigenous identity, location and sex. In some cases, CRVS systems include limited information about other socioeconomic variables that are useful for inequality monitoring, such as education level, literacy and occupation. They may also include identifiers, such as municipality of residence, that can be linked with sources of data on other dimensions of inequality, presenting expanded opportunities for inequality monitoring.

Beyond the high-income countries of the Americas, Europe and the Western Pacific, the quality of national CRVS systems is highly variable and is often in need of improvement (22). In many countries, CRVS data are lacking altogether due to large gaps in registration of vital events, lack of adequate resources, weak data collection systems, or incomplete legislative bases for requiring registration. Even where registration exists, the data may be of low quality, with missing information or significant biases, and therefore insufficient to serve their basic purposes. Evaluations of CRVS data quality may entail assessing the coverage of the CRVS system and completeness of data to determine their usability for monitoring.

Strengthening the role of the health sector in collecting birth and death information is an important part of establishing more complete and reliable CRVS data within populations (Box 12.4), which in turn will enhance the quality of inequality monitoring. For recording cause-of-death information, the use of the WHO International Form of Medical Certificate of Cause of Death is recommended for comparable and standardized data collection. This helps to ensure the underlying cause of death is reported in a reliable and systematic fashion (23). Standardized International Classification of Diseases (ICD) coding for causes of deaths is instrumental for the production of standardized, comparable and reliable statistics for use in health inequality monitoring. Strengthening the capacity of physicians in ICD-compliant medical certification of deaths is crucial for the collection of reliable data about causes of death.

BOX 12.4. Strengthening the role of the health sector in CRVS systems

With its network of services and unique access to populations at critical life stages, the health sector has a role in leading, contributing to and strengthening CRVS systems (16) WHO and UNICEF have developed guidelines to more effectively mobilize the health sector to support CRVS systems. The WHO civil registration and vital statistics strategic implementation plan 2021–2025 emphasizes strong leadership in the health sector, building of local capacity, and inclusion of marginalized populations to ensure no one is left behind (20).

The use of digital technology offers an unprecedented opportunity to improve the efficiency and accuracy of notifications for CRVS systems, and the subsequent processes of registering, aggregating and linking information. Internet access and mobile networks have enabled more timely and complete reporting of data from remote locations. In addition, online open-source platforms have made guidance surrounding data collection and the implementation of CRVS systems more widely accessible. Implementing national identification systems facilitates interoperability of CRVS systems with health and other administrative databases to deliver better services and reduce identity theft and fraud (24).

Censuses

A census is an official enumeration of a population, with systematic data collection from all members of the population. Many countries conduct national population and household censuses every 10 years (or in some cases, every five years). Additional censuses are sometimes conducted at subnational levels. Censuses provide essential information on population characteristics including age, economic status, ethnicity or race, geographical area, household composition and size, marital status and sex. Censuses are a comprehensive source of statistical information for economic and social development planning and administration. Various methodologies have been developed to conduct censuses (Box 12.5).

BOX 12.5. Census methodologies

Traditional census methodology involves the active collection of information from individuals and households on a range of topics at a specified time. Data collection, which may be done through long- and short-form questionnaires, occurs in a specified enumeration area over a short period of time. This ensures data collection is universal – covering all members of the population – and simultaneous. Short forms contain questions intended for universal coverage. Long forms, collecting more detailed information, are distributed to a sample of the population. Another common design involves the completion of a medium-length form by all members of the population. Censuses conducted through this traditional method have fewer and less complex data adjustments, because the raw data constitute all inputs (noting that non-response and the need for data validation, correction and imputation are part of traditional census processing).

Although most countries continue to use the traditional census approach, alternative census methodologies are gaining popularity (25). Alternative approaches may produce more frequent and timely statistics and require lower budgets and fewer inputs from the population, although they rely on more advanced technical capacities to process the data. These approaches include:

• rolling censuses, whereby information is collected through continuous cumulative surveys covering the whole country over a longer period of time;

• ad hoc sample surveys, conducted to provide information on topics not available from administrative sources, or for the purpose of making adjustments to poor-quality data in registers;

• existing sample surveys and registers, whereby information is collected from and linked across existing data sources.

Register-based censuses involve downloading information from a population register. Administrative censuses involve linking data from administrative sources to provide either a continuous or 10-year snapshot of the population. Both approaches provide additional dimensions of inequality not readily obtained from in-person questionnaires, although they do not capture items that can be obtained only through more traditional means, such as subjective health status.

Censuses are not usually health-focused, but they can include measures of health status and retrospective data on household or maternal mortality (although this is not common practice). Census data available for small geographical areas are useful for health-sector planning – for example, to determine access to health services or distribution of health workers. In settings where CRVS systems are lacking, the census may be used to gather information about births and deaths from proxy respondents. Like household surveys, censuses are often crucial sources of information on inequality in the social determinants of health, a vital part of inequality monitoring.

Strengths and limitations of census data

Major strengths of censuses for health inequality monitoring lie in their comprehensive coverage and collection of data on small geographical areas. The data collected through the census, although often lacking health data, can provide complete and accurate information pertaining to key demographic and socioeconomic dimensions of inequality. This information can serve as an important source of data about population sizes (useful, for example, for reweighting survey estimates), and socioeconomic information at the small area level (which can help to inform the selection of relevant dimensions for monitoring).

For the purposes of health inequality monitoring, the usefulness of the data may rely on the ability to link the data with other sources using small area identifiers such as postal codes or neighbourhood names. In some countries, census data contain identifying information at the individual level (e.g. through personal identity numbers), although access to and use of these data are highly restricted. More commonly, census data aggregated at the level of postal code or neighbourhood may be used to determine the average level of education or income for the area, which could then be linked with other sources of health information, such as primary care records, hospital episodes, vaccination records and mortality. Similarly, deprivation indices, which combine information across several socioeconomic dimensions of inequality to construct small-area level estimates, may be useful for health inequality monitoring (see Chapter 17). Data linkages often exist in high-income countries, but they may be lacking in many low- and middle-income countries.

Censuses are often scheduled to occur every 10 years and therefore the data may become out of date. In some settings, the timing of census data collection is not consistent, with delays due to reasons such as cost and complex logistics. Additionally, the period of data collection for a census may be lengthy, especially if financial or human resources are lacking. Census data can, however, be useful to “fill in” population numbers and make projections for the years between two consecutive censuses.

Use of population-based data sources for inequality monitoring

Population-based sources contain information that is representative of a base population, making them candidates for use in health inequality monitoring. Household health surveys, CRVS systems and censuses have different strengths and limitations, resulting in different applications for monitoring (Table 12.1).

TABLE 12.1. Considerations for using population-based data sources for health inequality monitoring

Data source	Strengths	Limitations	Opportunities to strengthen
Household health surveys	Surveys may include comprehensive information about health and dimensions of inequality Survey questions and methodologies can be adapted between survey rounds to address emerging issues Surveys are representative of national populations, regardless of whether they have contact with health or other administrative systems Repeating surveys over time generates comparable data useful for tracking changes in health inequalities Multicountry surveys that generate harmonized data across multiple settings facilitate benchmarking	Surveys may not be representative of smaller population subgroups or geographical areas Surveys may produce point estimates with high levels of uncertainty or that are subject to data suppression Surveys may be subject to sampling and non-sampling errors Surveys may be conducted infrequently, and data may become obsolete	Repeat surveys on a regular basis Increase the sample size of minority groups to ensure sufficient representation across subgroups Use reweighting to account for under-enumeration and response bias Harmonize questions across countries to facilitate benchmarking (e.g. through use of global frameworks to define health indicators and inequality dimensions)
CRVS systems	Designed to contain comprehensive, timely data about births, deaths and cause of death Routinely record information that enables disaggregation by age, place of residence or sex	Functioning CRVS systems require a high level of cooperation, coordination and investment across government agencies (or clear legal responsibility and funding of a single agency) and the health sector, which may be weak or lacking CRVS systems tend to lack information about socioeconomic inequality dimensions	Expand civil registration coverage to entire population (e.g. progressing from sentinel to sample to full registration) Build capacity for use of standardized international instruments to record cause-of-death data and ICD coding Expand collection of data on dimensions of inequality Collect information about personal or small-area identifiers to enable linkages with other data sources
Census	Data cover the entire population and provide accurate denominator counts, including by population subgroup Identifiers at small geographical levels, where available, are useful for linking with data from other sources	Health information tends to be limited Data collection occurs infrequently (usually every 5 or 10 years), and therefore may not be timely	Collect information about individual or small-area identifiers to enable linkages with other data sources Consider rolling, register-based or administrative censuses to improve frequency of data collection

CRVS, civil registration and vital statistics; ICD, International Classification of Diseases.

A first consideration is the availability of disaggregated data. Health data from household health surveys can typically be disaggregated according to multiple inequality dimensions, although the ability to disaggregate to small subgroups or geographical areas may be limited by small survey sample sizes. The possibility to meaningfully disaggregate CRVS data, however, tends to be very limited. Although the possibility exists, CRVS systems, even where highly functional, do not often include information on socioeconomic dimensions of inequality. CRVS systems are a rich source of data pertaining to births and deaths, but the scope of health information is narrower than that contained in household health surveys. The ability to link across data sources (see Chapter 15) provides expanded opportunities for the disaggregation of CRVS data. Censuses, with limited or no health data, tend to be rich sources of data about dimensions of inequality and the social determinants of health (such as economic status, education level or housing). In countries where census information is more limited, these data may not be directly useful for inequality monitoring, unless they are linked with another source through an individual or small-area identifier. In other countries, however, extensive inequality monitoring of health and health determinants using census data in conjunction with other data sources is common.

A further consideration for the use of these sources in inequality monitoring pertains to coverage. Although these data sources are aimed at representing or including the entire population, in reality they may fall short of this aim in different ways. Censuses often undercount certain vulnerable populations. In the United States of America, for example, migrants, homeless people, people from the LGBTQI+ community, children in foster care and people living with a disability are among the populations at risk of being missed (26). If not addressed through corrective measures, the implication of these gaps may be exacerbated further in household surveys when survey sampling frames are derived from census data.

Census data have a role in supporting the quality and use of other data sources. For example, information derived from the census is crucial for the design of household health surveys. It helps to ensure survey samples are designed to be representative of the entire population. In settings where other health data sources such as CRVS systems are weak or have incomplete coverage, the census may include questions about recent births and deaths. This information, collected through secondhand proxies such as parents or children, helps to correct for underreporting. Even cause of death, when evident, has sometimes been included in censuses, although these data are often of poor quality unless standardized verbal autopsy questionnaires are used.

Census data can also be used to determine accurate denominator estimates, which is applicable for the use of institution-based data to calculate rates or coverage. For example, although data from institution-based sources may contain information about the number of people who use a particular health service that they need (and this may be available disaggregated by age, location, sex and other dimensions), census data can be used to estimate the total population of people who need that service, including people who did not use the service. For more information on institution-based data sources, see Chapter 13.

Census data and household surveys are essential for ascertaining the distribution of social determinants of health across different social groups, which is crucial for monitoring inequality.

Data quality is an important consideration for the use of any data source for inequality monitoring. The adoption of standardized approaches, definitions and tools can greatly enhance the reliability and accuracy of measurements and support great comparability between populations. For example, using globally standardized definitions and criteria to define health indicators and inequality dimensions in household health surveys enables the practice of benchmarking. Adopting standard instruments and coding for recording cause of death is recommended for all CRVS systems.

The timeliness of population-based data sources is variable, owing to their different designs and overarching purposes. CRVS systems are designed to collect data on a continuous basis, while household health surveys and censuses are typically conducted as snapshots of the population, repeated on a recurring basis. These schedules impact the frequency with which inequality monitoring can be repeated.

References

1. O’Donnell O, Van Doorslaer E, Wagstaff A, Lindelow M. Analyzing health equity using household survey data: a guide to techniques and their implementation. Washington, DC: World Bank; 2008 (https://openknowledge.worldbank.org/entities/publication/8c581d2b-ea86-56f4-8e9d-fbde5419bc2a , accessed 9 August 2024).

2. WHO study on global ageing and adult health (SAGE). Geneva: World Health Organization; 2024 (https://www.who.int/data/data-collection-tools/study-on-global-ageing-and-adult-health, accessed 15 May 2024).

3. SHARE: survey of health, ageing and retirement in Europe. Munich: SHARE-ERIC; 2024 (https://share-eric.eu/, accessed 3 June 2024)

4. Demographic and Health Surveys Program. MIS overview. Rockville, MD: United States Agency for International Development (https://dhsprogram.com/methodology/survey-types/mis.cfm, accessed 15 May 2024).

5. STEPwise approach to NCD risk factor surveillance (STEPS). Geneva: World Health Organization (https://www.who.int/teams/noncommunicable-diseases/surveillance/systems-tools/steps, accessed 15 May 2024).

6. NCD microdata repository. Geneva: World Health Organization. (https://extranet.who.int/ncdsmicrodata/index.php/home, accessed 15 May 2024).

7. Global reproductive health surveys. Atlanta, GA: Centers for Disease Control and Prevention (https://www.cdc.gov/global-reproductive-health/php/surveys/index.html, accessed 15 May 2024).

8. Reproductive health surveys (RHS). Seattle, WA: Institute for Health Metrics and Evaluation (https://ghdx.healthdata.org/series/reproductive-health-survey-rhs, accessed 15 May 2024).

9. Performance monitoring for action. Baltimore, MD: Johns Hopkins Bloomberg School of Public Health (https://www.pmadata.org/, accessed 15 May 2024).

10. Demographic and Health Surveys Program. The DHS program. Rockville, MD: United States Agency for International Development (https://dhsprogram.com/, accessed 15 May 2024).

11. UNICEF MICS. New York: United Nations Children’s Fund (https://mics.unicef.org/, accessed 15 May 2024).

12. Global adult tobacco survey. Geneva: World Health Organization (https://www.who.int/teams/noncommunicable-diseases/surveillance/systems-tools/global-adult-tobacco-survey, accessed 15 May 2024).

13. Global youth tobacco survey. Geneva: World Health Organization (https://www.who.int/teams/noncommunicable-diseases/surveillance/systems-tools/global-youth-tobacco-survey, accessed 15 May 2024).

14. Demographic and Health Surveys Program. New optional modules for DHS-8 available. Rockville, MD: United States Agency for International Development (https://blog.dhsprogram.com/new-optional-modules-for-dhs-8-available/, accessed 15 May 2024).

15. Health Inequality Data Repository. Geneva: World Health Organization (https://www.who.int/data/inequality-monitor/data, accessed 15 May 2024).

16. World Health Organization, United Nations Children’s Fund. Health sector contributions towards improving the civil registration of births and deaths in low-income countries: guidance for health sector managers, civil registrars and development partners. Geneva: World Health Organization; 2021 (https://iris.who.int/handle/10665/341911, accessed 15 May 2024).

17. Principles and recommendations for a vital statistics system, revision 3. New York: United Nations Statistics Division; 2014 (https://unstats.un.org/unsd/demographic-social/Standards-and-Methods/files/Principles_and_Recommendations/CRVS/M19Rev3-E.pdf, accessed 15 May 2024).

18. Handbook on civil registration and vital statistics systems: management, operation and maintenance, revision 1. New York: United Nations Department of Economic and Social Affairs; 2021 (https://mdgs.un.org/unsd/demographic-social/Standards-and-Methods/files/Handbooks/crvs/crvs-mgt-E.pdf, accessed 4 June 2024).

19. Health Metrics Network. Framework and standards for country health information systems, 2nd edition. Geneva: World Health Organization; 2008 (https://iris.who.int/handle/10665/43872, accessed 15 May 2024).

20. WHO civil registration and vital statistics strategic implementation plan 2021–2025. Geneva: World Health Organization; 2021 (https://iris.who.int/handle/10665/342847, accessed 15 May 2024).

21. Issifou I, Pewitt J. On the importance of monitoring inequality in life expectancy. New York: United Nations Department of Economic and Social Affairs; 2022 (https://social.desa.un.org/sites/default/files/publications/2023-03/PB_145.pdf, accessed 15 May 2024).

22. World health statistics 2017: monitoring health for the SDGs, sustainable development goals. Geneva: World Health Organization; 2017 (https://iris.who.int/handle/10665/255336, accessed 15 May 2024).

23. Reporting cause of death. Geneva: World Health Organization (https://www.who.int/standards/classifications/classification-of-diseases/cause-of-death, accessed 22 June 2024).

24. Mitra RG. Linking national ID and CRVS systems: an imperative for inclusive development. Ottawa: Centre of Excellence for CRVS Systems, International Development Research Centre; 2019 (http://hdl.handle.net/10625/57766, accessed 3 December 2024).

25. Principles and recommendations for population and housing censuses, revision 3. New York: United Nations Department of Economic and Social Affairs; 2017 (https://unstats.un.org/unsd/publication/SeriesM/Series_M67Rev3en.pdf, accessed 15 May 2024).

26. Counting every voice: understanding hard-to-count and historically undercounted populations. Washington, DC: United States Census Bureau; 2023 (https://www.census.gov/newsroom/blogs/random-samplings/2023/10/understanding-undercounted-populations.html, accessed 15 May 2024).