According to CDC Guidelines “a public health surveillance system that is representative accurately describes the occurrence of a health-related event over time and its distribution in the population by place and person … to generalise findings from surveillance data to the population at large, the data from a public health surveillance system should accurately reflect the characteristics of the health-related event under surveillance”89. The key issue is to determine whether the current trachoma surveillance system is representative of the actual occurrence of trachoma in Australia. In so doing it is important to identify population subgroups that might be systematically excluded from the reporting system. Such groups may have been excluded based on age; stage of disease (trichiasis); geographical location and resource inefficiencies. Other factors that impact on representativeness are the calculation of rates (choice of denominator), the monitoring of risk factors for trachoma and data quality.

4.8.1 Population subgroups

The WHO recommends screening children aged 1-9 years old for trachoma in endemic regions. Current practice in Australia is to screen school children, most of whom are aged 5-14 years. It is generally accepted that this approach underestimates trachoma prevalence in 1-9 year olds as pre-school children carry the bulk of the community’s chlamydial load90. The NTSRU reports focus on Aboriginal children aged 1-9 years, complying with the CDNA guidelines. However, health services also collect data on children aged 10-14 and these data are only presented in a few of the NTSRU annual report tables. Data are predominately collected from children aged 5-9. According to the CDNA Guidelines for organisational and patient/family acceptance reasons it is much easier to screen school aged children, even though this approach generates data that are likely to underestimate trachoma prevalence.

Also, as we have already highlighted, there are issues associated with screening all the communities identified ‘at risk’. This problem is best illustrated by considering the data for SA. In 2008, only 16% (11 out of 72) of communities identified ‘at risk’ were visited and 365 out of 9,218 children (4%) were screened. In addition to this low community coverage, stakeholders suggested that some communities not identified as ‘at risk’ also have instances of trachoma. Thus it is fair to conclude that the surveillance system in SA is unlikely to be representative of trachoma occurrence in that state. Through information obtained from stakeholder consultations, it is suggested that the reasons for the low level of screening are largely due to resource limitations. top of page

The performance indicators presented in the National Framework for the Delivery of Trachoma Control Programsinclude two key measures regarding screening coverage in ‘at risk’ communities:
  • 80% of 5-9 year old children residing within ‘at risk’ communities screened for trachoma;
  • 90% of ‘at risk’ communities screened at least once every twelve months until prevalence of active trachoma is <5% for five consecutive years.
Based on data presented in Table 3.1, 51% of ‘at risk’ communities were screened in 2008 with wide variations between jurisdictions and regions; and 23% of children aged 5-9 in the ‘at risk’ communities were screened, again with wide variations between jurisdictions and regions. This result indicates that the levels of community coverage and screening coverage across the nation are well below current performance benchmarks set in the National Framework. It is unlikely, given such low coverage levels, that the current prevalence rate is representative of trachoma occurrence in ‘at risk’ communities.

As, as previously discussed, the recent National Indigenous Eye Health Survey (Taylor et al91) produced estimates of the rates of trachoma in very remote communities that were similar to the school based screening program in the NT, SA and WA, suggesting that the current surveillance system produces reasonable estimates of trachoma prevalence in high risk populations in those jurisdictions. However, there are apparent differences when the data from both sources are reviewed at the level of each region and the survey did find trachoma in communities that are not currently covered by the surveillance program; hence it is difficult to be confident about the surveillance system results.

Another component of the trachoma surveillance system is the collection of data on trichiasis in older Indigenous people. According to CDNA guidelines in areas where trachoma is endemic Aboriginal and Torres Strait Islander peoples aged 40-54 years should be screened every two years and those aged 55+ should be screened annually. Based on our consultations, there is currently no systematic screening for trichiasis in the NT and SA, rather, screening takes place on an opportunistic basis usually as part of ophthalmology outreach clinics. WA has a more systematic approach with trichiasis screening integrated into the influenza vaccination program. For 2008, data on trichiasis were reported for 26% (62 out of 235) of communities ‘at risk’ where 4% of the population at risk was examined (1,407 out of 34,603)92. It is unlikely that the screened communities are representative of the older Indigenous population, although there are no data available to test this proposition. Consequently, it is difficult to know whether the estimate of trachoma prevalence produced by the national surveillance system is representative.

Findings: With respect to the representativeness of the national surveillance system the evaluation team finds:
  • the current system is not likely to be highly representative of the prevalence of trachoma in 1-9 year old children (the WHO recommended target group) as:
    • only school age children are screened;
    • not all at risk communities are screened; and
    • it is known that there is trachoma in communities that are not screened.
  • the current system is not likely to be highly representative of the prevalence of trichiasis in older Aboriginals as:
    • there is no systematic approach to screening (except in WA); and
    • only a very small proportion of at risk communities are screened.
  • the benchmark coverage rate of 80% (as provided for in the funding and performance agreements between the Commonwealth and jurisdictions) together with a minimum number of children screened (dependent on the estimated number of children in the community) in ‘at risk’ communities should be adopted to ensure that screening in a community is sufficient to provide a reasonable estimate of prevalence.top of page

4.8.2 Calculation of rates

For many health-related events under surveillance, the proper analysis and interpretation of the data require the calculation of rates . The choice of an appropriate denominator for the rate calculation needs to be carefully considered to ensure an accurate representation of the health-related event over time, and by place and person. Three denominators are used in the reporting of the surveillance system data by the NTSRU that measure:
  • community coverage (number of communities screened);
  • screening coverage (number of children examined as a proportion of total children); and
  • prevalence of active trachoma, by age group.
Community coverage is calculated using the number of communities that were screened as a proportion of those that were identified by each jurisdiction as ‘at risk’. The denominator is thus the number of identified ‘at risk’ communities. As already highlighted the determination of communities ‘at risk’ is problematic across all states and territories. Our consultations show that ‘at risk’ communities are determined using historical knowledge and educated guess-work. One method of achieving a more methodological selection approach would be to formalise the criteria through which ‘at risk’ communities are identified and collect relevant information concerning those criteria. The criteria could be related to the factors known to be associated with trachoma prevalence. For example, criteria could include:
  • location – communities located in remote and very remote regions;
  • overcrowding – for example, the level of overcrowding revealed by the Community Housing and Infrastructure Needs Survey or more recent local information;
  • other environmental conditions data – again either derived from the Community Housing and Infrastructure Needs Survey or more recent local information;
  • results of previous screening activities; and
  • other local information e.g. from health care services, optometrists or ophthalmologists.
As the 2008 National Indigenous Eye Health Survey identified trachoma in populations where no active trachoma screening program exists, it is possible that the trachoma surveillance system is missing a number of communities that could be trachoma endemic. This finding suggests that the current surveillance system should be extended to additional communities (at least on a once-off basis) and supplemented by periodically conducting a national screening survey using a stratified sample.

Currently, the preferred method of calculating screening coverage is to use Australian Bureau of Statistics (ABS) population data to identify the number of children in ‘at risk’ communities. There are problems with ABS data due to difficulties in identifying Indigenous people, their high mobility and general issues associated with estimating small populations. Thus, it is considered that a better estimate of the population in a community on a screening day is obtained by examining school enrolment lists or local population data for the communities. In particular, stakeholders advised that a number of local primary health care services have client lists and that local community health nurses have knowledge of the population residing within the community at the time of screening.

As might be expected, the estimate of the population of children in the community varies depending on the source used. The NTSRU use ABS data to account for children in ‘at risk’ communities that were not screened. In ‘at risk’ communities where screening was conducted the number of children is estimated from school enrolments although it is only a surrogate measure (e.g. due to high mobility). For these reasons, the denominator to be used for calculating the screening coverage has been a contentious issue, but the use of the school enrolment lists, supplemented wherever possible by the local knowledge of health and/or education professionals is probably the best available method.

The prevalence of active trachoma is reported using the number of children examined in the community as the denominator. Given the high rate of absenteeism in schools in some areas, (in many cases absentees reach 50% each day94) the calculated prevalence may not be an accurate estimate of actual trachoma prevalence in the community. Often follow-up of students missing on screening day is not possible due to limited resources. Therefore, in some communities the number of children screened is small which can result in imprecision in estimates of trachoma prevalence (i.e. one more or less child identified as positive makes a big difference to the prevalence rate). As has been argued, it is unlikely that the children attending school who are screened are representative of those that are absent. Improving follow-up activities with absent children would improve the quality of the trachoma prevalence estimate for the community. Some strategies are already in place to increase attendance at school on screening day and stakeholders advised that since the inception of the NTSRU, the use of promotional and educational tools has helped increase attendance.

Findings: With respect to the calculation of trachoma prevalence rates in the national surveillance system the evaluation team finds:
  • the calculation of prevalence rates would be improved by:
    • formalising the definition of ‘at risk’ communities, by identifying objective criteria and decision rules and collecting information in relation to those criteria as part of the surveillance system;
    • screening in communities identified as not ‘at risk’ in remote and very remote regions at least on a once off basis;
    • screening ‘at risk’ communities in other regions at least on a once off basis, with repeated screening for communities with prevalence of greater than 5%;
    • increasing the number of children examined in each community through better follow-up strategies and initiatives to increase attendance at schools on screening days;
    • supplementing the surveillance data by conducting a five yearly national survey using a random stratified cluster sample.
  • there are considerable difficulties in determining an appropriate denominator for calculating screening coverage, a problem that is not unique to trachoma surveillance but inherent in determining correct population statistics in Indigenous populations residing in remote areas; and
  • the use of school enrolments as the basis for estimating the number of children in screened communities appears to be the best available method, but this figure should be supplemented with analysis of ABS estimated populations and by drawing on the local knowledge of health and education professionals in the community.top of page

4.8.3 Measurement of risk factor behaviour

The CDC guidelines suggest that the measurement of risk factor behaviours (e.g. facial cleanliness) might enable the monitoring of important aspects in the development of a disease or other health-related event95. It is understood from our consultations that trachoma occurs more commonly in dry, dusty conditions and is associated with sub-optimal living conditions such as overcrowding, reduced availability, and use, of water (for washing hands, faces and clothing), inadequate waste disposal, and high numbers of flies. The current surveillance system appears to collect reasonable information on facial cleanliness. However no information is collected on local environmental conditions. Data are collected in free text form on the implementation of health promotion and environmental activities.

Better collection of these data would enable the monitoring of important aspects in the development of trachoma. For example, if both current environmental conditions and interventions were included in data analysed, it would be possible to track improvements in environmental conditions over time and whether they were correlated with a reduction in trachoma prevalence. Similarly if better data were collected on health promotion and environmental initiatives, the impact of these activities could potentially be evaluated.

However, it is unrealistic to expect staff involved with screening to collect systematic information on environment and housing conditions in a community. These data are systematically collected every five years through the Community Housing and Infrastructure Needs Survey (CHIN). The survey was last conducted in 2006 and will be carried out again in 2011. It collected information about all discrete Indigenous communities and Indigenous organisations that provide housing to Aboriginal and Torres Strait Islander peoples. Data collected in the CHIN survey includes:
  • dwelling conditions – the state of repair of houses in discrete Indigenous communities;
  • water supply (including main source and associated water restrictions);
  • the communities sewerage system, including any leaks or overflows; and
  • the state of rubbish collection.
Other more local information is likely to be available on changes in environmental conditions in intervening years (e.g. in numbers of houses in the community). The evaluation team believes that these data sources should be explored with a view to linking the results to the national trachoma data. It is accepted that only very limited data on environmental conditions can be collected as part of the screening process (staff will not have the time, expertise or tools) so data linkage is the key to enhancing the measurement of risk factors in the trachoma surveillance system.

Findings: With respect to the measurement of risk factor behaviour in the national surveillance system the evaluation team finds:
  • the current information on environmental conditions is limited and does not facilitate the evaluation of the impact of interventions on trachoma prevalence;
  • key housing and environmental indicators from the CHIN survey should be incorporated into the data collection process (this could be best achieved at the national level) and made available to jurisdictional coordinators for local use (e.g. in assessment of at risk communities and for communication with local communities); and
  • the trachoma surveillance system would be improved by redesigning the collection of data on risk factor behaviour (including health promotion/education interventions) to introduce codified data elements that are easier to analyse and correlate to trachoma prevalence data.top of page

89CDC (2001) Updated Guidelines for Evaluating Public Health Surveillance Systems.Recommendations and Reports July 27, 2001/50(RR13);1-35
90Department of Health and Ageing (2006). Guidelines for the public health management of trachoma in Australia, Communicable Disease Network Australia
91Taylor HR, Fox SS, Xie J, Dunn RA, Arnold AL, Keeffe JE (2010) The prevalence of trachoma in Australia: the National Indigenous Eye Health Survey. Med J Aust, 92(5):248-53.
92Tellis B, Fotis K, Dunn R, Keeffe J and Taylor H (2009). Trachoma surveillance report 2008, National Trachoma Surveillance and Reporting Unit, Centre for Eye Research Australia
93CDC (2001) Updated Guidelines for Evaluating Public Health Surveillance Systems. Recommendations and Reports July 27, 2001/50(RR13);1-35
94In WA it was reported absentees can reach 50% each day. In the NT absentees are also an issue, but it is reported this is becoming less of a problem since the Australian Government Emergency Intervention.
95CDC (2001) Updated Guidelines for Evaluating Public Health Surveillance Systems. Recommendations and Reports July 27, 2001/50(RR13);1-35