Trachoma, caused by the bacteria Chlamydia trachomatis, is the leading cause of infectious blindness worldwide.1 It is endemic in 55 countries and may affect as many as 84 million people worldwide.2 3 Trachoma is associated with poor personal and community hygiene and poverty, and is now largely confined to the developing world.4 Australia is the only developed nation where trachoma is endemic,5 with the burden of disease falling almost exclusively on the Aboriginal population.6 Trachoma continues to contribute significantly to ocular morbidity in Aboriginal and Torres Strait Islander populations in parts of the Northern Territory, South Australia and Western Australia.7 Blindness from trachoma is preventable by the implementation of the Surgery for trichiasis, Antibiotic therapy, Facial cleanliness and Environmental impact (SAFE) Strategy promoted by the World Health Organization (WHO).2 3 It combines surgery for trichiasis and repeated single doses of azithromycin with promotion of facial cleanliness and improvement of environmental conditions.8

Clinical examination for trachoma, in the form of the simplified WHO grading system, has proven to be learnt easily by local health workers and generally has a high level of reproducibility.9 In Australia, trachoma is generally diagnosed as part of a screening programme. In principle, school children in remote communities receive an annual health check that includes trachoma screening. This process has been patchy at best, with there being little consistent data collection on the prevalence of trachoma in recent years in Australia.7 10 11 Further, despite recommendations for biennial screening for trichiasis of people aged 40–54 years and annual screening of people aged ⩾55 years in areas where trachoma is or has been endemic,12 very little screening of older people for trichiasis has been conducted.6 11

Clinical assessment can be difficult and results inconsistent when conducted by poorly trained or inexperienced individuals.4 An alternative method is to take photographs for later assessment by an experienced examiner.13 14 We report on the comparison of these methods for the assessment of trachoma in Aboriginal people in remote communities.

METHODS

A prevalence study of trachoma eye disease was conducted in the Katherine region of the Northern Territory during a 5-week period over July and August 2007. Five Aboriginal communities and their outstations were included in the study.

A community listing of those believed to be currently living in each community was constructed using the medical clinic patient list, the council housing list and the local knowledge of the Aboriginal Health Workers (AHW) seconded from the clinics to assist with the project. People who were believed to be only temporarily away from the community were included in the listing.

All people, including both Aboriginal and non-Aboriginal, in each community were invited to undergo a clinical eye examination for trachoma. Clinical examination was conducted using 2.5× magnifying loupes and a torch. The majority of examinations (96%) were conducted by one person (examiner A) to minimise inter-observer variability. The examiner had previously undergone training in clinical assessment for trachoma. The remaining examinations were performed by examiner B.

Each person was examined systematically, with both eyes being examined and scored independently for trichiasis, corneal opacity, follicles, intense inflammation and scarring. A fine grading scheme for each of the five signs of trachoma was devised to provide increased information about the levels of each stage of trachoma within the individuals and across the communities (table 1). A grading of 3 was equivalent to the parameters of the WHO simplified grading scheme15 for each of the five signs of trachoma.

Digital photographs were taken by examiner A of the left eye of each person following their clinical assessment. A digital Nikon D40 camera with an 85 mm lens, extension ring and flash was used. Photographs of the everted upper tarsal were graded on a laptop computer. The photographs were independently assessed and graded by examiner B without knowledge of the age, sex or trachoma grading assigned by the prior clinical assessment of examiner A.

Photographs were re-examined by both examiners when the clinical grade or the photographic grade was 3 or greater. Both examiners regraded these photographs and agreed on a final grade. A total of 88 photographs were re-examined for follicles, 29 for inflammation and 91 for scarring. This was conducted in a masked fashion in that no details of previous scores were visible during the adjudication viewing of the photographs. The final adjudicated scores were compared with the original clinical grading.

Data were entered into a Microsoft® Access (XP Professional) database and analysed using Stata® for Windows (version 9.0; Stata Corp., College Station, Texas, USA; http://www.stata.com/company/contact.html). Weighted kappa analysis was used to determine the concordance between the two methods of assessment,16 with a weighting of 1.0 being applied for agreement, 0.7 for one step mismatch, 0.3 for two steps and 0.0 for three or more steps of mismatch.

Ethics approval was granted by the Human Research Ethics Committees of the Royal Victorian Ear and Eye Hospital, the Australian National University and the Northern Territory Government Department of Health & Communities Services and Menzies School of Health Research. Signed written consent was obtained for each person, with consent for children under 18 years of age being provided by a parent or guardian.

RESULTS

Community screening

A total of 1316 people from a possible 1545 were screened at the five communities, giving an overall examination rate of 85.2%. Ninety-eight per cent of those screened were Aboriginal Australians. Most of those screened (54.6%) were female and nearly one-third (31.6%, n = 415) were <10 years of age. The mean age of those screened was 23.3 years (range 2 weeks to 95 years). The largest proportion of unexamined people occurred for both sexes in the 35–44-year-old age group, followed by the 25–34- and 15–24-year-old age groups. The highest response rates for examination were found in the school-aged children and the elderly.

A total of 1266 persons had both clinical and photographic assessment conducted. Fifty people from one community outstation were assessed clinically but not photographed for logistical reasons, as the camera was not available that day. Of the 1266 people photographed, only 13 digital photographs were insufficiently clear to enable an assessment to be made.

Prevalence of the signs of trachoma

A total of 115 people were assessed clinically as having active trachoma, that is, follicles (TF) and/or inflammation (TI) in one or both eyes. The youngest child observed with active trachoma was just over 1 year old. The majority (89.6%, n = 103) were <15 years old

A total of 201 people were found to have clearly visible scarring (TS) of the tarsal conjunctiva in clinical assessment. The youngest person identified with scarring was 7 years old. Most people with scarring (87.6%, n = 176) were aged ⩾35 years. Of those people aged ⩾40 years, 61.5% (n = 158) had scarring.

Six people were identified with trichiasis (TT) and a seventh person indicated that they had previously had bilamellar tarsal rotation surgery to correct trichiasis. The mean age of the six people with unoperated trichiasis was 63.6 years old. The youngest person identified with trichiasis was 46 years old and had severe trichiasis affecting both eyes. Across the study population, the prevalence of unoperated trichiasis was 4.6 per 1000 people. No one was observed to have corneal opacity.

The age-specific clinical prevalence of each of the four signs of trachoma observed in this study is shown in table 2.

Comparison of clinical and photographic assessment

There was 79.7% agreement between clinical assessment and photographic assessment for the assessment of trachomatous follicles with a weighted kappa of 0.40 (table 3).

When the fine scores for follicles were translated into the WHO simplified score, TF, there was 96.1% agreement with a weighted kappa of 0.71.

There was 81.6% agreement between clinical assessment and photographic assessment for the assessment of trachomatous inflammation, with a weighted kappa of 0.22 (table 4).

When the fine scores for inflammation were translated into the WHO simplified scores, TI, there was 99.2% agreement, with a weighted kappa of 0.37.

There where 1254 photographs in which scarring could be assessed. There was 89.3% agreement between clinical assessment and photographic assessment for scarring, with a weighted kappa of 0.67 (table 5).

When the fine scores for scarring were converted into the WHO simplified grading, TS, there was 92.7% agreement, with a weighted kappa of 0.67.

Comparison of right eye to left eye by clinical assessment

The fine graded scores for clinical assessment for right eye and the left eye were compared and weighted kappa scores calculated. The agreement for trachomatous follicles between the right and the left eye from clinical assessment by the same person was 94.6% with a weighted kappa of 0.81. When the scores were converted to the WHO simplified grading system 42 people were scored as having unilateral TF and 69 had TF in both eyes.

The agreement for trachomatous inflammation between the right and the left eye from clinical assessment by the same person was 97.0%, with a weighted kappa of 0.78. When the scores were converted to the WHO simplified grading system two people were scored as having unilateral TI and two had TI in both eyes. There was 99.6% agreement between WHO simplified scores for right and left eyes for inflammation with a weighted kappa of 0.44.

The agreement for scarring between the right and the left eye from clinical assessment by the same person was 95.7% (weighted kappa 0.87). When the scores were converted to the WHO simplified grading system, TS, 45 people were scored as having unilateral scarring and 156 had TS in both eyes with 96.9% agreement (weighted kappa of 0.85).

DISCUSSION

This study found that both active trachoma in young people and scarring in older people remain a problem in Aboriginal communities in the Northern Territory. Overall, averaged across these communities active trachoma was at endemic level (>10%) in children <10 years of age. Two communities had a prevalence of active trachoma of >20% in children <10 years of age. The prevalence of active trachoma is similar to that reported 30 years ago in this region by the National Trachoma Eye Health Program.17 A disturbing finding was that the prevalence of trichiasis was found to be substantially higher than the “less than one case per 1000” target set by WHO for the elimination of blinding trachoma.18

Although antibiotic treatment has been provided by the local health services for the communities screened in this study, other communities in the region were neither screened nor treated at the time of this survey. There is therefore the risk of re-infection due to strong cultural and family ties and the mobility of people between communities. Sustained action across all communities in the region is required to ensure long-term management of trachoma, including a more systematic approach to screening, treatment and prevention. Implementation and resourcing of the guidelines for trachoma control developed by the Communicable Diseases Network Australia is required.12

Previous studies have sought to determine the validity of using photographic grading to assess for signs of trachoma using prints or slides, with mixed outcomes.13 14 19 West and Taylor13 in 1990 used slide photography and concluded that photographic documentation was valid and reproducible. In contrast, Solomon et al14 in 2006 reported only “fair to moderate” agreement between clinical and slide photographic assessment. In 2004 Emerson et al19 used a combination of print and digital photography and reported substantial agreement between clinical and photographic assessment. The present study sought to determine the usefulness of digital photography in a field environment compared with clinical assessment.

Overall these data demonstrate that there was quite good agreement between the clinical grading and the photographic grading as rated by kappa scores, with the greatest strength of association being for trachomatous follicles and trachomatous scarring.

However, compared with clinical grading, the photographic grading tended to relatively overgrade the presence of inflammation but not of follicles or scarring. Alternatively, these may have been under-called by clinical grading with the difficulties inherent in fieldwork. This difference may be a function of the larger computer photograph and the ability to spend considerably more time examining a stable image than is possible in clinical assessment of an inverted eyelid of a person.

These findings are in agreement with those of West and Taylor13 but are not consistent with the findings of Solomon et al.14 It is unclear why there is the disparity in the different studies, but the results may depend upon the clinical grading skills and the quality of the photographs being assessed, as well as the skill of the photo-grader.

Despite the possible limitations in terms of relative overgrading, the use of photographic grading does have some distinct benefits. Photographic grading can serve as a backup for inexperienced clinical examiners in situations where a more experienced examiner cannot be present. Photographic assessment provides objective monitoring, and thus minimises the risk of drift in grading that may occur over time, which is particularly important in longitudinal studies. The use of standardised photographic assessment conducted by one person provides an appropriate counterbalance for situations where there are multiple clinical graders.

The use of photography was readily accepted by the participants in the study. Despite the young age of many of the children no particular difficulties were encountered with either the clinical assessment or the photographing. The great benefit of using a digital camera was that the photographs could be checked quickly to determine if the image would be clear enough for later assessment and repeated as necessary.

It is also important to note that although the simplified grading system set by WHO as the criteria for assessing trachoma enables standardised assessment and ease of determination of who requires treatment, it does have limitations. The fine grading scheme as used in this study uncovered large numbers of people with “subclinical” signs for trachoma, particularly a grade 2 level. These people would be classified as normal by graders using the WHO simplified grading scheme. The use of a fine grading scheme is especially important in research studies that look for correlations of clinical disease with environment or genetic risk factors of the distribution of infection in a community when attempts are made to separate those with diseases from those who are “normal”. Clearly there are many people who do not meet the clear-cut definitions used in the WHO simplified grading who still have significant active disease and who would be misclassified as normal. Having said this, we recognise that the purpose behind the simplified grading was to have a simple and reproducible system for health workers to use in programmatic activities and we strongly support its continued use for this purpose.

In summary, this study found that photographic assessment is a valid technique with many benefits, particularly for situations where clinical assessment by a competent and confident person may not be possible. Awareness of the potential for relative systematic undergrading for scarring and overgrading for inflammation across a population will enable appropriate assessment of final scores.