More information about text formats
We apologise for stating that the paper by Radford et al. was based on a questionnaire reporting survey rather than active surveillance reporting (ASR) as set up by the British Ophthalmic Surveillance Unit (BOSU). The latter system relies on clinicians reporting monthly to a central office (BOSU). We note that BOSU is only responsible for collecting data and is not involved in publications by o...
We apologise for stating that the paper by Radford et al. was based on a questionnaire reporting survey rather than active surveillance reporting (ASR) as set up by the British Ophthalmic Surveillance Unit (BOSU). The latter system relies on clinicians reporting monthly to a central office (BOSU). We note that BOSU is only responsible for collecting data and is not involved in publications by others who use it. In contrast, we recommend that authors publishing data collected by BOSU should involve them as this may reduce the chance of under-ascertainment of cases used to calculate incidence figures. It may also lead to standardisation for calculating missing, unreported cases.
Foot & Stanford comment how Radford et al. estimated ascertainment through independent sources using hospital microbiology and pharmacy records. However, not all cases of Acanthamoeba keratitis (AK) are culture-positive. Anti-amoebic therapy such as Brolene(R) (propamidine eye drops 0.1%) has often been given prior to the patient's hospital attendance, while many laboratories have little skill in isolating this amoeba. If seen in private practice, a trial of therapy is often given without collecting a corneal scrape specimen and without NHS pharmacy prescription. In our opinion this approach is unreliable.
Foot & Stanford  maintain that 93% of 'reporting' ophthalmologists thought they had reported all cases of AK to BOSU but this does not mean that 93% of cases were reported to BOSU. What happened to those cases seen by the ophthalmologists who did not report to BOSU? The mean annual card return rate by region for the ASR of BOSU was 71%  so do they mean that 93% of 71% i.e. 66% of cases were reported? This figure (66%) is often quoted for the return of questionnaire reporting surveys (QRS) both in our experience, that of Foot & Stanford  as well as others. This suggests that the same two thirds of respondents reply whether by ASR or by a QRS!
We agree with Foot & Stanford that the incidence rate reported by a surveillance project must be considered as the minimum rate. Authors should heed their warning ! We also agree that it could be wrong to assume that the correction factor for the unreported cases should be directly proportional to the level of non-response and we suggested suitable ways to offer statistical correction. However, at the simplest level an adjustment can be made based on the level of non-response. If the study is large, then errors made from adjusting in this way will be minimal. A similar method of ascertainment adjustment has been used by researchers evaluating post-operative endophthalmitis based on BOSU reporting data.
Foot & Stanford suggest that the ASR scheme is best for epidemiological investigation of rare diseases. This may apply for some diseases but we disagree for infectious diseases. We believe that a prospective cohort study in a well-defined region, with extensive planning and discussion with health care professionals, is more reliable to identify the 'true' incidence rate. Patients with AK present in public and private ophthalmology practice, to private doctors, to optometrists and to contact lens-practitioners. Patients wearing contact lenses (CLW) can present early with swollen lids, epitheliopathy and conjunctival inflammation, as well as corneal nerve infiltration but without any form of ulcer. It is reasonable to remove the lens and treat with Brolene(R) which is sold as an over-the-counter product by non-hospital pharmacists without a prescription. Indeed, three million units are sold annually in the UK often for treating red eyes in CLW. Such cases will not be reported to BOSU but were reported in our cohort studies in Scotland  and Hong Kong  as we had frequent meetings with local Associations of Optometrists. Hospitals and private doctors were notified to acquire all possible community-based cases. In the Scottish study, 3 out of 14 cases of AK in CLW were recognised by optometrists in the city to whom they first presented.
The annual incidence of AK between 1997/99 was similar at 1 per 30,000 CLW for both Hong Kong using a cohort study and England using the ASR scheme, provided that there is correction for non-responding ophthalmologists. The English figures do not include community-based non-hospitalised cases so there is still some under-reporting. Hong Kong has similar contact lens practices to England with the exception that chlorine disinfection tablets were never marketed there. The incidence in the Scottish cohort study was four times higher at 1 per 6750 CLW due to frequent use of the chlorine tablet 'Softab'and tap water in 1994/956.
Is it still worthwhile continuing with the ASR of BOSU for cases of AK? The incidence rate has dropped due to the withdrawal of 'Softab' and the introduction of multi-purpose solutions that are both effective against Acanthamoeba and provide a sterile cleaning and disinfecting solution.[8,9] The use of daily-disposable and new extended-wear silicone hydrogel lenses will further reduce the incidence. The BOSU reporting scheme will identify some cases but, as Radford et al. reported, reliability was poor; some cases were duplicates while others did not fit the diagnostic criterion.
Cohort studies can be expensive but so is the annual cost of the BOSU! Which is better value? Both our cohort studies [6,7] cost approximately £80,000 each, but when completed, the cost stops. Essential, reliable information was gained that will be of value for the next 10 years. If the BOSU is to survive, it must constantly review the conditions that it includes in its ASR schemes. It should no longer include AK or other infections. The BOSU should also consider the advantages of a one-time QRS to alleviate the boredom of monthly card reporting. Full-time researchers prefer well-defined and well-planned prospective studies but those in full-time clinical practice may prefer active surveillance or questionnaire reporting but must be careful with their interpretion.
David V. Seal
Applied Vision Research Centre, City University, Northampton Square, London EC1V 0HB
Tara K. Beattie
Department of Vision Sciences, Glasgow Caledonian University, Cowcaddens Road, Glasgow, G4 0BA
Departments of Ophthalmology and Centre for Clinical Trials and Epidemiological Research (School of Public Health), Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong
(1) Radford CF, Minassian DC, Dart JKG. Acanthamoeba keratitis in England and Wales: Incidence, outcome and risk factors. Br J Ophthalmol 2002;86(5):536-42.
(2) Foot BG, Stanford MR, Rahi J, Thompson JR. The British Ophthalmological Surveillance Unit: An Evaluation of the First 3 Years. Eye 2003;17:9-15
(3) Foot BG Stanford MR. Questioning Questionnaires. Eye 2001;15(6):693-4.
(4) Schmitz S, Dick HB, Krummenauer F. Endophthalmitis in Cataract Surgery. Results of a German Survey. Ophthalmology 1999; 106 (10): 1869-76.
(5) Seal DV, Beattie TK, Tomlinson A, Fan D, Wong E. Acanthamoeba Keratitis. Br J Ophthalmol 2003;87(4): 516.
(6) Seal DV, Kirkness CM, Bennett HGB, Peterson M & Keratitis Study Group. Population-based cohort study of microbial keratitis in Scotland: incidence and features.
Contact Lens & Ant Eye 1999; 22: 49-57 & ibid 58-68..
(7) Lam D, Houang E, Lyon D, Fan D, Wong E, Seal D. Incidence and Risk Factors for Microbial Keratitis in Hong Kong: comparison with Europe and North America. Eye 2002; 16 (5): 608-618
(8) Stevenson RWW, Seal DV. Has the Introduction of Multi-purpose solutions contributed to a reduced Incidence of Acanthamoeba keratitis in contact lens wearers? Contact Lens & Ant Eye 1998; 21: 89-92.
(9) Beattie TK, Seal DV, Tomlinson A, McFayden AK, , Grimason A. Determination of Amoebicidal Activities of Multipurpose Contact Lens Solutions by Using a Most Probable Number Enumeration Technique. J Clin Microbiol 2003; 41: 2992 - 3000.
The comment of Seal et al. on the report by Radford et al. on the incidence of acanthamoeba keratitis highlighted a number of points of interest when interpreting the results of a surveillance project carried out through the British Ophthalmological Surveillance Unit (BOSU).
The correspondence describes the study as a questionnaire survey.
This is incorrect; cases were asc...
The correspondence describes the study as a questionnaire survey.
This is incorrect; cases were ascertained through prospective
based active surveillance. This methodology has been shown to be
in trials, identifying up to twice as many cases as passive methods [3,4]
and has been evaluated in ophthalmology. Furthermore, they also
that the BOSU has some authorship rights over the report. This is also
incorrect. The BOSU provides a methodological framework for assessing
incidence and clinical features of rare eye conditions. Cases are
identified through our surveillance system and notified to independent
research groups who contact the reporting ophthalmologist directly to
collect information about the reported case. The BOSU does not collect
information about the patients and as such does not have any claims to
ownership of the data or request to be listed as the author in any
Seal et al. do highlight the likelihood of underascertainment in
this study. This
a recognised source of potential bias and when possible the BOSU does
encourage researchers to ascertain cases through an additional
source to allow the incidence to be estimated using capture-recapture
analysis. In the event of this being unavailable they should make
to estimate ascertainment and identify possible sources of error.
et al did this through comparison with microbiological and pharmacy
records at selected centres.
The assertion by Seal et al. that the ascertainment for this study
the response rate achieved during the study period by the BOSU reporting system is inaccurate. Any incidence rate reported by a surveillance
project must be considered as a minimum rate and it is wrong to assume
that the correction should be proportional to the level of non-response.
Despite no reliable independent source to assess ascertainment existing
for this study, a subsequent survey of reporting ophthalmologists
indicated that they had reported 93% of cases of acanthamoeba keratitis
Case ascertainment achieved through the BOSU is not directly linked
to response rate, and can differ between conditions. Barriers to
participation in the surveillance scheme have been reported. In
addition, difficulties with the interpretation of the case definition,
management by non-consultant ophthalmologists or reluctance by
participating ophthalmologists to report certain conditions could be
relevant. Whilst we appreciate that some errors in estimation of
are inherent active surveillance through the BOSU continues to provide
best opportunity for the epidemiological investigation of rare eye
(1) Seal DV Beattie TK Tomlinson A Fan D Wong E, Acanthamoeba
Keratitis. Br J Ophthalmol 2003 87(4):516.
(2) Radford CF, Minassian DC, Dart JKG. Acanthamoeba keratitis in
and Wales: Incidence, outcome and risk factors. Br J Ophthalmol 2002; 86(5):536-42.
(3) Thacker SB Redmond S, Berkelman RL, A controlled trial of
surveillance strategies. Am J Prev Med 1986:2:345-350
(4) Vogt RL, LaRue D, Klaucke DN, Jillison DA, Comparison of an
passive surveillance system of primary care providers for hepetitis,
measles, rubella, and salmonellosis in Vermont. Am J Pub
(5) Foot BG, Stanford MR, Rahi J, Thompson JR. The British
Ophthalmological Surveillance Unit: An Evaluation of the First 3 Years. Eye 2003;17:9-15
(6) Foot BG Stanford MR. Questioning Questionnaires. Eye 2001;15(6):693-4.