Proyecto Banco del Migrante

Capture-recapture analysis is a recognised method for quantifying completeness o f a s c e r t a i n m e n t . A l t h o u g h previously used in the study o f congenital a n o m a l i e s , i t s application to the study o f ophthalmic disorders has not been described before. The technique was applied here in its two-source method which, to be valid, requires that certain conditions are fulfilled, namely that all cases are drawn from a closed population, are identifiable for matching, and have a uniform probability o f being ascertained by each source, which are independent o f each o t h e r . ( S e c t i o n 3.2.3.1, page 84)

In this study, by restricting the capture-recapture analysis to cases bom in 1995 or 1996 and detected in infancy, a subgroup effectively forming a birth cohort, was identified as the closed population o f interest.

Accurate matching was ensured by the availability o f more detailed information on all reported cases.

The issue o f uniform ascertainment probabilities was more complex. It was reasonably assumed for reporting by ophthalmologists, with the only probable exception being infants identified by paediatricians but not referred to

ophthalmologists for treatment and there were no such reports. The two infants who died before ophthalmic assessment could take place were still reported by paediatricians and added strength to the belief that, as intended, children with systemic disorders associated with cataract had been adequately ascertained by surveillance through the paediatric scheme. Thus a uniform probability o f identification by paediatricians, who are involved in both detection and

management o f infants with cataract, was likely to have existed but is difficult to prove.

Comparison o f cases reported through the ophthalmic and paediatric schemes showed that they were similar with respect to factors which might have influenced ascertainment: laterality, treatment undertaken and age at, and context of,

detection (in particular, detection through screening in infancy). Whilst the percentage o f notified cases with idiopathic cataract was similar in the two schemes, a higher percentage o f infants notified through the paediatric than the ophthalmic scheme had associated systemic disorders and the converse was true

for infants with isolated ocular disease. As it is recognised that some level o f variable catchability exists in any natural p o p u l a t i o n , t h e s e differences were not considered to be sufficient to preclude the use o f capture-recapture analysis.

Complete independence o f reporting sources is rare.^°° Consequently, although paediatricians and ophthalmologists notified their cases through entirely separate systems, the possibility that paediatricians did not report cases they knew had been notified by their ophthalmic colleagues, or the converse situation, cannot be entirely discounted. Thus some dependency between the two sources cannot be excluded. Had positive dependence in reporting existed, whereby a case was more likely to be identified in one system if also reported in another, this would result in an overestimate o f the completeness o f ascertainment^°^’^°^ An

assessment o f the order o f magnitude o f such an error can be made by assuming the estimated 95* centile o f the true number o f infants in the population to be correct: this would suggest that overall ascertainment had been 86% (161/187) rather than 92% (161/175). Conversely, had negative dependence o f sources existed, resulting in under-estimation o f ascertainment, o f generally o f much greater magnitude,^°^ this would imply that overall ascertainment had been higher than 92%. Neither positive or negative dependence, or the existence o f both, in this study can be excluded.

In the absence o f other data sources, and given the small number o f cases reported through the paediatric scheme alone, assumptions about uniform ascertainment

probabilities and source independence could not be tested statistically in a

meaningful way/^^ However, from knowledge o f the role o f the ophthalmologists and paediatricians in the UK in the management o f congenital and infantile

cataract, as well as o f the two reporting schemes, it is reasonable to suggest that use o f capture-recapture analysis to estimate ascertainment o f infants was valid. Thus the derived estimate o f detection rates o f the two surveillance schemes are considered accurate.

The high level o f ascertainment achieved by the ophthalmic scheme, established within five months specifically for this study, concurs with the experiences of other investigators^''’ and demonstrates that effective surveillance can be achieved by clinicians without prior experience o f the methods. The high ascertainment is likely to reflect the motivation of participating ophthalmologists who, as members o f the study collaborative group, were committed to the longer-term assessment of outcome in the study cohort. Their participation in the surveillance scheme is also likely to have been encouraged by the high level o f communication maintained with them throughout the study p e r i o d , a s well as by the combination and flexibility o f methods o f reporting, which are reported to be important

components o f successful surveillance schemes.

However, although effective, had there been sole reliance on the ophthalmic scheme, 5% o f all cases would have been missed, which re-emphasises the value o f using multiple appropriate sources, wherever available, to enhance

ascertainment in epidemiological studies o f rare disorders.’’^ Furthermore, without two sources, estimates o f the completeness o f ascertainment would not have been possible.

As surveillance has not been widely applied in ophthalmological

research,*^^’’’^’'^^’*®^ experience o f its use in this study, as a disorder-specific and versatile reporting model, strengthens the initiative by others to establish a new national active surveillance scheme in ophthalmology, run by the British Ophthalmological Surveillance Unit (BOSU),'^^’’’’ and may also have implications for its methods.

Flexibility is recognised to be an important attribute o f successful surveillance s c h e m e s a n d in the present study, allowing respondents to report cases by means other then the formal notification system improved ascertainment. Thus flexibility and adaptability o f the notification system may be important to the success o f specific studies undertaken through the BOSU scheme in the future. Recognition o f passive reporting within a system that it intended to be active is also important in evaluating and interpreting compliance and ascertainment.

Specific surveillance systems are reported to be more accurate, which is supported by the findings o f the present study. Thus, it is possible to suggest that those rare ophthalmic disorders managed by a small number o f sub-specialists on the BOSU reporting base, may be more efficiently and thoroughly ascertained by

surveying only the relevant subset o f respondents. These and other possible benefits, such as reduced administrative requirements and costs, would have to be weighed against the potential disadvantage of ascertainment bias which might be introduced by failing to identity any eligible cases under the care o f

ophthalmologists who individually only manage a very small number o f affected individuals, for example one case per year.

Surveillance through the paediatric scheme improved ascertainment overall, and in particular, helped to identify children with systemic disorders and those not undergoing treatment. Thus one main purpose of surveying through this scheme was served. However ascertainment through this scheme was low despite high compliance with the notification system. Assuming that null reports were accurate, this might have reflected other factors which it would be important to address in future studies o f ocular disorders using this scheme.

Since its establishment over 10 years ago, the scheme has facilitated the concurrent surveillance o f a number o f uncommon d i s o r d e r s . W h e r e specifically assessed, the level o f case ascertainment has varied with the disorder being studied.^^’’’^’'^^ for example, higher ascertainment was reported for children with d i a b e t e s , a n d for babies with vertically transmitted both disorders in which paediatricians have a greater role in primary management, than for cataract in this study. Differences in ascertainment are likely therefore to reflect the pattern o f presentation and management o f the disorder s t u d i e d , a s well as

the composition and completeness o f the reporting base with respect to

subspecialty interests.^’'^ As previously discussed, all these factors may have been relevant in the present study, as both community paediatricians and

neonatologists, reported to be under-represented in the paediatric reporting base^’'* might be expected to be involved in the detection o f children with cataract.

As prevalent cases, identified by either scheme, were ineligible, it is probable that the temporal sequence o f detection and referral o f some cases was o f greatest importance to the ascertainment o f cataract by the paediatric scheme. Cases first detected within the study period but subsequently examined by paediatricians outside it, would not have been notified through both schemes and this accounted for 28% o f cases correctly notified by paediatricians.

It is possible that all the notified but unconfirmed cases were eligible, and thus their exclusion reduced the sensitivity (detection rate) o f the scheme. Difficulties in tracing notified cases subsequently have been reported previously'^^’’^* and in the present study might have been avoided by providing paediatricians with data collection questionnaires at the start o f the study, just as ophthalmologists were given data proformas.

Lastly, it is possible that some cases known to paediatricians may not have been reported, which is supported by the experiences o f other investigators using this s c h e m e . I n this context, paediatricians may have been less motivated to report

cases than ophthalmologists. Their response might have been encouraged by providing regular individual feedback on study progress, in a similar way as to ophthalmologists. This would be important to consider in any future studies through this scheme o f disorders not primarily managed by paediatricians.

As ophthalmic disorders have not previously been surveyed through this, or other national schemes in paediatrics’’* the findings regarding ascertainment o f

congenital and infantile cataract may be relevant to similar studies in the future. A number o f measures can be taken to maximise the specificity and sensitivity of reporting o f ocular disorders in such schemes. These include the use o f an easily applied case definition, an understanding o f the clinical workload o f different groups o f respondents, the characterisation of the role o f paediatricians in the management o f affected children, together with the temporal sequence o f their involvement, and the maintenance o f appropriate communication to improve the motivation o f respondents to notify cases and provide data about them.