Health statistics and information systems

Report: First Consultation of the High Level Advisory Panel on Health Statistics

Notes from the Consultation

1. Comparative advantages of WHO

WHO's work in health statistics needs to build upon its comparative advantages, which include:

  • its constitutional and legitimate link with its Member States, which mandates and facilitates reporting of health data, and is essential to ensure effective use of health information for decision-making;
  • its convening power to reach consensus and facilitate harmonization both at regional and country levels with lead partners in the field of health;
  • its ability to set standards and norms, and develop and harmonize methods and tools for health information, in collaboration with relevant partners.

2. Setting priorities is crucial

  • Given the resource constraints, there is a need for WHO to focus on a limited number of topics; the High Level Advisory Panel should focus on advising WHO on priorities, monitoring activities, coordinating mechanisms across organizations and, equally importantly, making recommendations on what WHO should not do.
  • WHO's priorities are reporting on mortality, morbidity, health status, service coverage and risk factors.
  • WHO's role in producing financial statistics and health systems metrics may also needs to be discussed by the Panel but will require additional experts to participate in such a meeting.
  • There should be clarity about the frequency with which specific health statistics are produced. Some need to be produced on an annual basis (e.g. coverage of key interventions such as immunization coverage), others – such as composite measures of health – much less frequently, driven by data availability.
  • The construction of composite measures of health is to be limited in frequency not because they are intrinsically less important, but rather because they require more effort, and the data upon which they are based in many countries are weak. Reliable cause-specific mortality data are lacking in many countries and this needs to be addressed as a priority. The morbidity and health status data on which they the composite measures are also based are even weaker and this needs to be addressed as a second priority. If there were good data of both sorts in any given country, the work involved in constructing composite measures is then marginal, and they could be more frequent.
  • ICD-10 and related normative work is an important priority. There is a need to develop, disseminate and promote a short list of causes of death for use in resource-limited settings.
  • WHO should also consider investing more in painting the picture of health in the next ten years.

3. Demand-driven supply of health statistics

  • New global health initiatives such as the Global Alliance for Vaccines and Immunization (GAVI) have created strong demands for more frequent production of health statistics. Such demands should be considered as an opportunity to invest in strengthening data collection and health statistics generation.
  • However, some of the requests are difficult to respond to and WHO should consult advisory groups and make explicit the grading of evidence if the 'new' estimates are not driven by new data.
  • WHO should also give due consideration to the demand of global health initiatives such as Global Fund and GAVI for independent assessment of the quality of data, e.g. in the context of disbursement of funds based on output statistics, by developing tools and methods to audit data quality. Currently, only one such tool is available (on immunization coverage).
  • WHO should ensure that World Health Assembly resolutions which involve target setting reflect realistic data collection where applicable.

4. Data collection in countries

  • The Panel focused on the importance of population health data in developing countries. WHO should take a leadership role in developing strategies for country data collection, capacity building and coordination of such activities across different partners.
  • WHO has collected mortality data from developed and developing countries for decades and should continue to work with countries to do so.
  • Supporting countries to collect cause-specific mortality by age and sex should be the first priority, and WHO has given too little attention to promoting vital registration systems as a source of mortality data in high mortality countries, especially in Africa. Even information from selected areas would be useful. There is evidence that a continuous sample of representative demographic surveillance systems is a useful source of information on cause-specific mortality.
  • Household surveys are the major source of information on health issues related to the living population, including morbidity, health status, health service coverage and risk factor prevalence. The inclusion of biological and clinical data collection in surveys can greatly expand the role of household surveys in monitoring morbidity and risk factor prevalence (health examination surveys). In addition, household surveys continue to be a major source of information on levels of under-five mortality and have the potential to assess levels of adult mortality including maternal mortality. It is also the main source on disparities. Better harmonization among donors that demand survey data results is however needed.
  • The potential use of census data to assess mortality should be considered in collaboration with key partners, as part of the UN Decade of the 2010 round of censuses.
  • There is a demand for reliable subnational information on coverage trends (e.g. immunization, antenatal care, integrated management of childhood illnesses), and WHO needs to promote subnational survey tools that can be implemented at relatively low cost but result in data of sufficient quality to monitor trends and validate or adjust clinic-based data.
  • WHO needs to develop strategies to effectively direct and mobilize Global Fund and other funds in countries to address the dearth of data on the major diseases and build capacity to collect and analyse such data in the future.
  • The Health Metrics Network (HMN) provides an opportunity to strengthen specific data collection platforms, particularly by leveraging country-level international and domestic resources for monitoring and evaluation. WHO and its partners should advocate for HMN support for investment in data collection platforms, notably vital registration and also appropriate household survey systems that generate national and subnational health data.

5. Producing estimates of health statistics

  • WHO needs to be transparent in producing health statistics. Key aspects are ensuring comparability between countries, the ability to establish time series, and the communication of uncertainty ranges especially when it comes to providing country-level estimates.
  • Grading of evidence is essential as follows: no or very limited country data available, based on modelling of data from other countries or historical populations; some data available, but availability and quality are clearly inadequate; (partially) adequate information available.
  • The proposed WHO four-step process for the production of regional and global estimates at headquarters – accessible database, transparent methods, independent advisory group and overall consistency through EIP clearance – is endorsed.
  • External reference groups for epidemiology of major diseases have been useful for ensuring accuracy and transparency of estimates and there is a critical need for a reference group on adult health epidemiology.
  • WHO has a major role in compiling and archiving country-reported data and making them accessible to the wider public (e.g. time-series data on mortality by age, sex and cause). Careful assessment of reported data is essential to ensure comparability and quality. Furthermore, WHO should make major efforts to support the in-country application of estimation procedures, which includes simplifying tools and methods to the extent possible and building national capacity.
  • WHO needs a more integrated approach to setting standards and producing health statistics across departments. The Panel requested WHO to put in place a cross-cutting collaboration of technical departments through the proposed World Health Statistics Programme and to strengthen its capacity at both regional and country levels.

6. Mortality and causes of death

  • WHO should take the following steps to improve the availability of country-level mortality data by age, sex and cause ("count the dead"):
  • Continue and expand its core function of collecting (secondary) data on mortality by age, sex and cause, and work with countries to make better use of and enhance access to existing (primary) data in close collaboration with regional offices.
  • Establish a WHO adult mortality / health reference group to address the methodological issues such as model life tables, the use of biased data, improving cause-of-death model, management of uncertainty of mortality estimates, and how to handle ill-defined and garbage codes.
  • Advocate for investment in country (primary) data collection through appropriate data collection methods, with a particular focus on vital registration/demographic surveillance systems.
  • Produce country-level estimates (see 9. Country-level estimates and consultation). If properly released, country estimates could help countries to identify data gaps and facilitate better data collection.

7. Morbidity, health status, health service coverage and risk factors

  • Estimates of morbidity, health status, health service coverage and risk factors are an essential part of health statistics and should be a priority of WHO's work in this area.
  • WHO plays a key role in the surveillance of major infectious diseases and an adequate response to a limited number of acute infectious diseases that require immediate action, is essential. The reporting of some conditions, however, is of limited value because of poor quality and of limited value for global responses. A critical reappraisal of such reporting practices is necessary.
  • The reporting of several other diseases and conditions such as cancer (through national registries) or tuberculosis, is essential for making estimates and monitoring trends over time. Reported service coverage is subject to bias but may still play a role in making estimates if investments are made to improve the data quality.
  • Population-based surveys can provide key information on selected morbidity, health status, and risk factors, but there is a need for further work in this area (set research agenda).

8. Composite measures of health

  • Summary measures consisted of four major activities:
    • conceptual framework,
    • input data - primary data collection – both mortality and morbidity / health status,
    • estimation to produce the best age- and sex- and cause-specific mortality, morbidity / health status estimates for the country’s entire population, and
    • construction of summary measures.
  • The production of summary measures (disability-adjusted life years, healthy life expectancy and comparative risk assessment) is important from the global public health point of view, but should not be done frequently: updating with a 5- or 10- year interval is useful. If however new data availability allows more frequent estimation of e.g. HALE, updates should be done more frequently.
  • WHO should invest in country data and estimates on mortality, morbidity, health status and risk factor prevalence which would serve as essential inputs to the modelling. Mortality is a major driving force of changes in global burden of disease estimates, but it would be much better to have data on morbidity / health status too.
  • WHO should focus on the overall framework and country data on mortality, morbidity, health status, and risk factor prevalence that go into the modelling, rather than investing in the modelling itself. This includes clear articulation of the strengths and weaknesses of the summary measures and its value in national health information systems. WHO should oversee the entire process of summary measure assessment within a statistical framework, to ensure a degree of legitimacy with a clear indication of weaknesses and strengths and their value in national health information systems.
  • The required modelling may be done better by academic or other institutions outside WHO. WHO should collaborate with such institutions, contribute to the production of estimates, and require transparency and reproducibility of the estimates at the country level.

9. Country level estimates and consultation

  • There is a great demand for (frequent) statistics on cause-specific mortality, morbidity, health status, health service coverage, risk factors, including global health initiatives, planning tools such as marginal budgeting for bottlenecks (MBB), or demand to evaluate interventions.
  • Several disease and programme areas have implemented good models for estimation that meet the basic standards for the production of country-level health statistics – database, transparent methods that are accessible to countries, advisory groups or consistent consultation with academia, and effective country consultation. In several programmes, working in partnership has been extremely useful.
  • Modelling, in the sense of extrapolation or interpolation, does not capture in-country heterogeneity. Yet there is considerable demand for such information, which has major implications for programmes. WHO should advocate and support the development and implementation of data collection through such means as demographic surveillance systems and surveys on specific areas such as health services coverage.
  • Release of model-based estimates of, for example, cause of death pattern at the country level can be carried out if the following are in place:
    • clearance of 4-step procedure and accompanying grading of evidence (see 5. Producing estimates of health statistics);
    • country consultation;
    • emphasis on and clear communication of the uncertainty associated with the estimates;
    • clear understanding to countries of the methods used to obtain the estimates.
  • Country-level statistical and epidemiological capacity to adjust for biases, synthesize and analyse data is weak in most developing countries. There are several ways to strengthen analytical capacity – closer collaboration with the statistical constituency in countries, investment in user-friendly tools at the country level for data analysis and enhancing use for decision making. The Health Metrics Network should contribute to this important element of its country health information systems strengthening framework and strategy.
  • WHO needs to work closely with national statistical offices.

10. Research agenda

  • It is critical for WHO to consider what data should be collected for the research agenda. WHO should recommend which indicators should be collected and how to collect them through surveys or other platforms.
  • The functions of the Panel should be linked to Advisory Committee on Health Research.
  • Some ideas for research: mortality data ad estimates, comparative risk assessment and the use of counterfactuals, assessment of health information systems

11. Functions and objectives of the Panel

  • WHO will decide on whether to make the Panel ad-hoc or standing. A standing panel has long-term advantages: more power and more continuity. It will learn and gain experience over time.
  • Membership needs to be revised to make the Panel to ensure that it is representative of all the various health areas. It should also include representatives from countries and the major players in health statistics.
  • WHO Secretariat should revise the draft terms of reference accordingly.