Verteporfin photodynamic therapy for neovascular age-related ...

DOI: 10.3310/hta16060Health Technology Assessment 2012; Vol. 16: No. 669was not administered at the recommended frequency or under the conditions determined bythe licensing trials. Even minor diminution of efficacy could have resulted in VPDT becomingcost-ineffective given the borderline benefit of VPDT. Thus, providers and purchasers of healthcare need to be aware that the characteristics of treatment in routine practice may be substantiallydifferent to the treatment recommendations made on the basis of licensing trials. Our findingsalso highlight that commercial trials may recommend more treatment than necessary or ata treatment frequency that is not deliverable across eligible populations. This fact is also ofimportance to researchers when designing pragmatic phase 3 trials. 62,63Visual function and health-related quality of lifeThe large sample in the VPDT study gave us adequate power to test the secondary objectiveson HRQoL relationships and adaptation to vision loss over a 2-year period, even though theproportion of patients with data for more than three visits was small. Approximately half ofthe centres provided data on a second measure of vision, namely CS, and also administeredstructured and validated instruments to ascertain visual functioning, HRQoL and resourceutilisation. Thus, only a subset of participants in the VPDT study contributed the data for theanalysis of the secondary outcomes as they were necessarily a selected sample and had to beattending one of these centres. These limitations may have led to selection in terms of socioeconomicstatus or age. However, previously published work shows that these factors do notinfluence HRQoL or visual functioning. 18,31Our analyses of the associations between BCVA and HRQoL identified three main features:■■■■■■Health-related quality of life (SF-36 and the NEIVFQ) decreased with deteriorating visualfunction (BCVA or CS in the better-seeing eye) over a wide range of visual function.The relationship between visual function and HRQoL measures was not sigmoid but tendedto plateau at low levels of visual function.The gradient of the relationships did not change over time up to 2 years after first treatment.Health-related quality of life is a ‘whole-patient’ outcome, which is dependent on patients’binocular visual experience on a day-to-day basis using their habitual correction (if any) and intheir customary environment. Previous studies have demonstrated weak relationships betweendeclining visual function and generic HRQoL instruments and stronger relationships withvisual functioning instruments such as the NEIVFQ. However, the association between themost commonly used surrogate marker for visual function, that is BCVA, and the NEIVFQ is atbest moderate, with the majority of the variation remaining unexplained. Reasons for the lackof a strong association include that (a) BCVA itself is a psychophysical test and is influenced bypatient factors; (b) BCVA subserves only foveal function and does not reflect the more complexaspects of overall vision such as reading text, depth perception, movement detection, colour andcontrast processing and field of vision; and (c) self-reported HRQoL has an in-built variabilitywhich is dependent on the respondent’s mood and emotional status.Clinical tests to obtain estimates of additional or more global aspects of visual function are noteasily measured in routine practice and were beyond the scope of the VPDT study. Instead,we attempted to collect presenting binocular BCVA with the participant wearing the habitualcorrection, if any. We reasoned that this measure would better reflect the usual state of visualfunctioning and, thus, the HRQoL response. However, one-third of the data describingpresenting binocular BCVA were missing and many of the remaining data were collected in avariety of formats, making meaningful interpretation of the findings difficult. Therefore, likeprevious researchers, we were forced to select BCVA in the better-seeing eye as the proxy forbinocular visual performance. 18,31,64 Validation of the BCVA data showed that BCVA had beencollected in a far more robust and reproducible manner.© Queen’s Printer and Controller of HMSO 2012. This work was produced by Reeves et al. under the terms of a commissioning contract issued by theSecretary of State for Health.