NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Paediatric HIV Surveillance Among Infants and Children Less Than 18 Years of Age. Geneva: World Health Organization; 2013.

Cover of Paediatric HIV Surveillance Among Infants and Children Less Than 18 Years of Age

Paediatric HIV Surveillance Among Infants and Children Less Than 18 Years of Age.

Show details

4Quality of the paediatric surveillance system

The ability to make reasonable estimates of HIV prevalence among the paediatric population using any surveillance method relies heavily on the coverage and quality of data, and the quality of diagnostics. The data and laboratory systems involved in paediatric surveillance should be regularly evaluated to ensure that quality standards and procedures are in place and being followed.

4.1. Data system quality

This section discusses overall issues relating to data coverage and quality relevant to all surveillance methods described in this document, broadly categorized as survey-based surveillance methods (immunization clinic surveys, population-based surveys and special surveys) and routine data-based methods (case reporting, mortality data).

4.1.1. Data coverage

Data coverage is the proportion of the target population that is accessed by a particular surveillance method. It may be an indicator of how representative the surveillance data are of the target population. To assess the representativeness of surveillance data, you would need to identify the potential selection biases involved in a particular surveillance method. Thus, additional variables that may predict inclusion in a particular surveillance method should be collected.

For example, if PMTCT programme data are used to calculate the number of HIV-exposed infants, one must consider the extent to which PMTCT services exist and are used in that setting. Questions to be asked and described in any analysis or dissemination of results include the following:

  • What percentage of HIV-positive pregnant women are accessing services?
  • What percentage of these women accept HIV testing?
  • What factors predict differential testing rates for subgroups of pregnant women (for example, those with a known HIV status)?

4.1.2. Data quality

Data quality has four general components: completeness, timeliness, consistency and accuracy.

Completeness refers to the extent to which data are reported fully for any given surveillance method. This may be assessed at multiple levels:

  • The number of observed paediatric infections divided by the number of expected paediatric cases through modelling or other prevalence estimates
  • For survey-based approaches, the survey and/or testing response rate among children
  • For routine data-based methods, the proportion of facilities that are reporting paediatric cases
  • The full reporting of paediatric surveillance data elements or the completion of data fields (forms, surveys, testing).

Timeliness refers to the speed with which data are available through a surveillance system. Surveillance data are timely when they are current and when the information is made available rapidly; timeliness can be assessed at facility, subnational and national levels. For example, timeliness for a paediatric case-based surveillance system can be calculated from the time of diagnosis of paediatric HIV infection to the time of receipt of the paediatric case report form at the national level. In general, the frequency and timeliness of data from survey-based surveillance will be less than that of routine data-based surveillance, where data are collected on a more continuous basis. For surveillance data to be useful for implementation of prevention, care and treatment programmes, health staff must receive timely feedback on the results of the surveillance. Timeliness can also be assessed by the rate at which surveillance data are incorporated into policy planning and programme implementation.

Consistency or reliability of surveillance data depends on the use of standard protocols and consistent procedures at each site over time. The use of standard protocols and procedures allows the data to be collected consistently, regardless of who does the data collection or how often data are collected.

Accuracy is also known as validity, meaning that the data correctly measure what they are intended to measure. Accurate data have minimal errors.

  • For survey-based surveillance methods, errors include interviewer bias, or sampling or data collection errors.
  • For routine data-based surveillance methods, errors include overcounting or undercounting cases, and transcription errors (whether the information on the abstraction form matches information in the patient record at the health facility).

4.2. Laboratory quality

Quality laboratory testing is critical to any surveillance programme. Poor laboratory results can lead to inaccurate estimates, misdiagnoses, incorrect treatment or delays in treatment. Laboratory testing in the context of paediatric HIV surveillance must include the appropriate testing methodologies for children of different ages, as well as quality assurance of the testing algorithms. In addition, blood sample collection may enable further surveillance activities among paediatric populations, including CD4 or viral load testing, drug resistance and ART exposure. Therefore, it is recommended that laboratory quality be considered an integral and vital component of paediatric HIV surveillance programmes.

Evaluation of laboratory quality should include the following:

  • Availability and regular maintenance of written standard operating procedures for all procedures and staff training
  • Appropriate sample collection (e.g. serum, plasma, whole venous or capillary blood), processing and transport
  • Appropriate storage of reagents and biological samples
  • Use of validated or approved diagnostic test kits, reagents and controls
  • Use of test kits and reagents within their manufacturer-provided shelf-life
  • Inclusion of positive and negative controls within each batch of tests
  • Calibration and routine maintenance of laboratory equipment
  • Safety guidelines and procedures in place
  • Participation and satisfactory performance in an external quality assurance programme (EQAP) or interlaboratory comparison
  • Proper recording and evaluation of individual and summary results, and quality control of data
  • Maintenance of patient confidentiality in all laboratory records and throughout sample processing.

4.3. Validate paediatric surveillance data against other sources

Given the general lack of paediatric surveillance data in most settings, it is useful to validate existing surveillance data through alternative methods. In this section, we describe how to use mathematical models to verify the results of a surveillance system. This is especially important among new and developing surveillance systems such as paediatric HIV surveillance systems. It is important to note that models cannot replace surveillance systems. Models rely on assumptions and are likely to change as the HIV epidemic is better understood. For these reasons, models should be used cautiously and should themselves be validated based on empirical data.

Models can be used to supplement and validate results from surveillance systems. Examples of how models have been used to validate adult surveillance data include validating the decline in HIV prevalence in Zimbabwe and its link to changes in behaviour(40) and the decline in mortality in Botswana after widespread ART coverage.(41) Examples of how models have been used to validate paediatric surveillance are not available.

4.3.1. How models work

There are a large variety of models that help to understand epidemics and demographic changes. Mathematical models simply describe a system and the dynamic change in that system with different inputs and assumptions. In the case of HIV, HIV prevalence and epidemic changes are often modelled, given HIV risk and exposure in the population and variations in transmission modes and transmission rates.

4.3.2. Currently used software modelling packages

Many countries use the Spectrum computer package to estimate the impact of their HIV epidemic. Spectrum can describe the HIV epidemic at the country and subnational levels. Spectrum is a discrete compartmental model, which means that it progresses individuals by age group through different disease states. More information on the model and how it works can be found elsewhere.(42,43) Additional information on Spectrum is provided below.

The Asian Epidemic Model (AEM) is used to determine national HIV prevalence and track new HIV infections in a country to determine where new infections are arising. Similar to Spectrum, the AEM supplements existing HIV surveillance systems to inform policy and programmes.(44)

Outputs from Spectrum are available for various paediatric age groups and include the following:

  • The number of new HIV infections in the previous year
  • The current number of prevalent HIV infections
  • The number of children who need co-trimoxazole prophylaxis
  • The number of HIV-related deaths
  • The number of children needing its coverage
  • The number of pregnant women living with HIV.

To estimate the number of children living with HIV, the Spectrum model incorporates assumptions about MTCT rates and survival of HIV-infected children. Country-specific inputs are also used, such as the estimated HIV prevalence in women (aged 15–49 years), and coverage of PMTCT and ART programmes.

The accuracy of Spectrum estimates depends on the quality of the data included in Spectrum to develop the model, as well as the assumptions and parameters used in the model. Spectrum is revised routinely based on new information resulting from research on HIV. In addition, countries improve their models whenever country files are updated with additional surveillance data.

Currently, estimates from Spectrum of HIV prevalence among young children are not informed by direct paediatric surveillance data. In addition, Spectrum-generated paediatric estimates can be unreliable for concentrated or low-level epidemics. In these settings, there is usually very little information on fertility among the populations with the highest levels of HIV prevalence (sex workers, partners of clients of sex workers), making it difficult to estimate the number of children infected with HIV through MTCT. Thus, paediatric surveillance activities are particularly important in concentrated or low-level epidemics.

4.4.3. Using models to validate or supplement surveillance

In countries where there is no functioning paediatric surveillance system, models can be used to get a rough understanding of the potential number of children who might be HIV-positive. Spectrum currently estimates only the number of children (younger than 15 years of age) infected through MTCT. Other routes of transmission are not captured. The results from models can roughly estimate the trends of new infections and the distribution of existing infections to determine the magnitude of HIV among children.

In countries where some surveillance data are available for children, models can provide figures to compare against surveillance estimates to determine quality and/or validity. For example: if the only available paediatric surveillance data come from case reports, a programme manager might assume that this is an undercount of the actual number of children living with HIV, since some children may not come into contact with testing or care and treatment services. In such settings, a model might be able to estimate the number of expected child infections, thus providing an adjustment for the case-reporting system. Similarly, if a surveillance system is operational in only a limited number of areas, a programme manager might assume that the results are not representative. Models could be used to extrapolate how many HIV-positive children living in those areas were not captured in the surveillance system.

Models can also be used to supplement surveillance data. Often, countries have fairly good systems to monitor HIV prevalence but do not have good systems to measure deaths due to AIDS. Mortality surveillance is especially challenging among young children who might not yet have been recorded in national vital registration systems. Models can estimate HIV-associated paediatric mortality. However, mortality rates from models will only be as good as the assumptions used in the models.

4.5. Triangulate data

Paediatric surveillance data should be triangulated with other data sources to validate and better understand observed results. According to the WHO guidelines on data triangulation,(45) “triangulation can be broadly defined as the synthesis and integration of data from multiple sources through collection, examination, comparison and interpretation. By first collecting and then comparing multiple datasets with each other, triangulation helps to counteract threats to the validity of each data source.” Triangulation can be applied to better understand and explain HIV trends for M&E purposes, as well as for resource allocation and advocacy. More importantly for paediatric HIV surveillance, triangulation can help redirect interventions. For further information on data triangulation and its processes, challenges and end-points, please refer to additional triangulation guidelines.(46,47)

Copyright © World Health Organization 2013.

All rights reserved. Publications of the World Health Organization are available on the WHO web site ( or can be purchased from WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel.: +41 22 791 3264; fax: +41 22 791 4857; e-mail: tni.ohw@sredrokoob).

Requests for permission to reproduce or translate WHO publications –whether for sale or for non-commercial distribution– should be addressed to WHO Press through the WHO web site (

Bookshelf ID: NBK158972
PubReader format: click here to try


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...