72 Discussion of resultsA further concern that has been previously expressed is that the utilities <strong>for</strong> the HUI3 havebeen derived from a power trans<strong>for</strong>mation of values from a visual analogue scale, 72 ratherthan by direct valuation with choice-based methods such as the standard gamble (used <strong>for</strong> theSF-6D) or the time trade-off (used <strong>for</strong> the EQ-5D). Our approach of using the SF-6D followsthe recommendations of policy-makers such as NICE. 15 There<strong>for</strong>e, it is not surprising that ourfindings are consistent with previous studies that have used the EQ-5D. 64In terms of an internationally recognised measure of HRQoL appropriately based on societalpreferences, the gradient of the decrease in HRQoL with deteriorating visual function is small.The estimated gain in HRQoL (utility) from VPDT is about 0.02 (in terms of BCVA, a differenceof about 11 letters after 2 years, assuming that only the best-seeing eye is being treated), andfrom ranibizumab is about 0.04 (a difference of about 21 letters after 2 years, 73,74 under the sameassumption). Gains in utility (over varying time horizons) <strong>for</strong> other common interventions <strong>for</strong>chronic conditions (Table 23) show that the utility gain associated with VPDT is relatively smallcompared with other competing interventions. 75,76 These utilities measured over the appropriatetime horizon, and combined with relative effects on life years gained, translate into QALYs andin<strong>for</strong>m health policy decisions.Cost of illness and resource utilisationThe VPDT cohort study, unlike most other studies used to estimate cost-effectiveness, collecteddata concurrently on both resource utilisation and HRQoL. Thus, it was able to report on thecost-effectiveness of VPDT versus BSC under the assumption that BSC involved scheduled visitsto the ophthalmology clinic to monitor patients’ vision and no other treatment.The main empirical finding from the CEA is that the costs of providing VPDT <strong>for</strong> patientsincluded in the UK VPDT cohort study were relatively high compared with the projected QALYTABLE 23 Utility gains of VPDT compared with other common interventionsIntervention a Utility gain b Duration of follow-up c Measure used dCataract surgery 0.03 3 months EQ-5DGroin hernia repair 0.06 3 months EQ-5DTotal hip replacement 0.42 6 months EQ-5DVaricose vein surgery 0.10 3 months EQ-5DTotal knee replacement 0.31 6 months EQ-5DCoronary artery bypass grafting 0.21 6 years EQ-5DVPDT e Year 1: 0.009; year 2: 0.012; total: 0.021 2 years SF-6Da The utility gains shown were selected as comparators because the data were collected when the interventions were provided in usualhealth-care clinical settings. Utilities <strong>for</strong> five surgical interventions (cataract, hernia, varicose veins, knee replacement surgery and total hipreplacement) were obtained from the Patient Reported Outcome Measures (PROMs II) project. 75 Utilities be<strong>for</strong>e and after coronary arterybypass grafting were elicited from patients rated clinically appropriate <strong>for</strong> both bypass surgery and percutaneous man<strong>age</strong>ment in theAppropriateness of Coronary Revascularisation (ACRE) study. 76b All of the utilities were measured be<strong>for</strong>e and after the intervention.c The utility gains are reported <strong>for</strong> different durations of follow-up after an intervention, so they are not directly comparable. In general, onewould expect the utility gain to be larger with a longer duration of follow-up. Note that the 2-year utility gain from VPDT is still lower than theutility gains achieved over a shorter durations <strong>for</strong> the interventions studied in the PROMs II project.d The PROMs II and ACRE studies measured utility gains using the EQ-5D, 75,76 whereas the VPDT cohort study measured utility using the SF-6D.The EQ-5D and SF-6D have been shown to differ when used to report extreme health states; 70 the EQ-5D appears unable to discriminatehealth states close to full health (ceiling effect), whereas the SF-6D seems unable to discriminate health states close to zero (floor effect).However, on aver<strong>age</strong> the two instruments generate utilities that are very similar each other, with a mean difference of only 0.05, 70 so thedifferences shown in the table cannot be explained by the use of different utility measures.e The estimate <strong>for</strong> <strong>photodynamic</strong> <strong>therapy</strong> assumes that the better-seeing eye is being treated (see text).
DOI: 10.3310/hta16060Health Technology Assessment 2012; Vol. 16: No. 673gain. The incremental cost per QALY was £170,000 over 2 years <strong>for</strong> the base case. Even if thegain from the intervention was extrapolated over 5 years, the incremental cost per QALY wasstill approximately £100,000, five times higher than the threshold that NICE uses to identifyinterventions that are ‘relatively cost-effective’. 15 The cost-effectiveness ratios reported in previousCEA of VPDT <strong>for</strong> nAMD have ranged from US$30,000 to US$250,000 (£20,000 to £166,667assuming an exchange rate of US$1.5 to the UK pound). 26,28 Unlike previous CEA of VPDT<strong>for</strong> patients with predominantly classic CNV secondary to nAMD, we were able to include allappropriate costs and account <strong>for</strong> HRQoL associated with vision loss. This resulted in improvedpatient-centred estimates that showed that VPDT is unlikely to be cost-effective.There are three other features of our CEA that are relevant to future CEA of interventions <strong>for</strong>nAMD. Firstly, the study used data on the use of a treatment in routine practice rather than datacollected in accordance with a trial protocol. Despite the much lower treatment frequency, weobserved similar visual outcomes to those observed in the TAP trials. In the case of our CEA,it was more appropriate to use the lower treatment frequency observed in the VPDT cohortstudy, and previous CEAs that used treatment frequencies from the TAP trials overstated theincremental costs of the VPDT intervention compared with BSC. 19,26,77 The lesson <strong>for</strong> future CEAis that treatment intensities observed in licensing trials may overestimate the treatment intensityused in routine practice and, hence, overestimate the costs of treatment in the usual care setting.Second, previous CEAs either excluded costs associated with declining vision or estimatedthese costs based on expert opinion. 19,26,28,77–79 By contrast, by collecting patient-level data onHSS resource use and BCVA, our CEA was able to incorporate the cost of declining vision. Ourdata showed that the HSS costs <strong>for</strong> patients with nAMD were low (e.g. a mean of £320 <strong>for</strong> BSCgroup in year 1), and hence the reduction in these costs after VPDT was relatively small (£151over 2 years). The costs were lower than those reported by a recent observational study assessingthe economic burden of nAMD by self-reported use from 400 patients across five differentcountries 23 and from studies of Medicare costs based on claims data. 80,81 However, these studieswere based on aggregated costing approaches, which tend to overstate costs. The morbiditycosts observed in the VPDT cohort study relied entirely on patient recall and there<strong>for</strong>e mayhave under-represented the true cost, and also they may reflect the relatively poor availability oflow-vision services in the UK; previous studies have found a similarly low use of vision-<strong>related</strong>services in the UK. 82,83 However, this low use of vision services has been reported in othercountries that have used patient-level data, which suggests that the current findings may be morewidely applicable. 24 The lesson <strong>for</strong> future CEA is that the source and robustness of data describingcosts associated with declining vision need to appraised with care.The third feature of our CEA that is relevant to future CEA concerns the importance of usingHRQoL measures based on preference weights from the general population rather than patientswith nAMD.A final issue <strong>for</strong> future CEAs which emerged from the VPDT study but which was notincorporated into our CEA was the relationship between BCVA and HRQoL. The regressionmodels which investigated this relationship found that the rate of change in HRQoL with varyingBCVA was not influenced by whether the better- or the worse-seeing eye was being treated.However, a key assumption in this and other CEA is that it is always the better-seeing eye thatis being treated, as the HRQoL gain is ‘credited’ <strong>for</strong> all treated eyes. Unless policy-makers rulethat worse-seeing or ‘first’ eyes should not be treated (an option considered but rejected byNICE during its deliberations prior to issuing its technology appraisal 6 ), the worse-seeing eyewill be treated in a proportion of patients (48% in the VPDT cohort study). There<strong>for</strong>e, the CEAreported here (and other CEAs) has overstated the QALY gain and the cost-effectiveness of© 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 <strong>for</strong> Health.
- Page 5 and 6:
DOI: 10.3310/hta16060Health Technol
- Page 12 and 13:
xExecutive summaryretreat on criter
- Page 14 and 15:
xiiExecutive summaryof data, with i
- Page 16 and 17:
2 Introductionon-going or new clini
- Page 18 and 19:
4 Introductionof VPDT in routine cl
- Page 21 and 22:
DOI: 10.3310/hta16060Health Technol
- Page 23 and 24:
DOI: 10.3310/hta16060Health Technol
- Page 25 and 26:
DOI: 10.3310/hta16060Health Technol
- Page 27 and 28:
DOI: 10.3310/hta16060Health Technol
- Page 29 and 30:
DOI: 10.3310/hta16060Health Technol
- Page 31:
DOI: 10.3310/hta16060Health Technol
- Page 34 and 35:
20 Key changes to the protocolThese
- Page 36 and 37: 22 Key changes to the protocolCentr
- Page 38 and 39: 24 Key changes to the protocolADVER
- Page 40 and 41: 26 Key changes to the protocolPlan
- Page 42 and 43: 28 Key changes to the protocolClass
- Page 44 and 45: 30 Key changes to the protocolFor i
- Page 46 and 47: 32 Key changes to the protocolIn ad
- Page 48 and 49: 34 Key changes to the protocolsites
- Page 52 and 53: 38 Results (1) - study cohortTABLE
- Page 54 and 55: 40 Results (1) - study cohortTABLE
- Page 56 and 57: 42 Results (1) - study cohortTABLE
- Page 58 and 59: 44 Results (1) - study cohortTABLE
- Page 60 and 61: 46 Results (2) - objectives A, B, C
- Page 62 and 63: 48 Results (2) - objectives A, B, C
- Page 64 and 65: 50 Results (2) - objectives A, B, C
- Page 66 and 67: 52 Results (2) - objectives A, B, C
- Page 69 and 70: DOI: 10.3310/hta16060Health Technol
- Page 71 and 72: DOI: 10.3310/hta16060Health Technol
- Page 73 and 74: DOI: 10.3310/hta16060Health Technol
- Page 75 and 76: DOI: 10.3310/hta16060Health Technol
- Page 77: DOI: 10.3310/hta16060Health Technol
- Page 80 and 81: 66 Discussion of resultsStrengths a
- Page 82 and 83: 68 Discussion of resultsThe VPDT st
- Page 84 and 85: 70 Discussion of resultsOur analyse
- Page 88 and 89: 74 Discussion of resultsVPDT. It is
- Page 91 and 92: DOI: 10.3310/hta16060Health Technol
- Page 93 and 94: DOI: 10.3310/hta16060Health Technol
- Page 95: DOI: 10.3310/hta16060Health Technol
- Page 98 and 99: 84 References16. Stelmack J. Qualit
- Page 100 and 101: 86 References49. Guymer RH, Chiu AW
- Page 102 and 103: 88 References83. Margrain TH. Minim
- Page 104 and 105: 90 Appendix 1THE VERTEPORFIN PHOTOD
- Page 106 and 107: 92 Appendix 11. Overview of Manual
- Page 108 and 109: 94 Appendix 1And, if also collectin
- Page 110 and 111: 96 Appendix 1more patients in the v
- Page 112 and 113: 98 Appendix 12.4 Limitations of the
- Page 114 and 115: 100 Appendix 1Box 2 Key advantages
- Page 116 and 117: 102 Appendix 14. Study population4.
- Page 118 and 119: 104 Appendix 15. Recruitment to the
- Page 120 and 121: 106 Appendix 16. Background data co
- Page 122 and 123: 108 Appendix 1Table 1: Schedule of
- Page 124 and 125: 110 Appendix 1The Verteporfin Photo
- Page 126 and 127: 112 Appendix 1• Additional clinic
- Page 128 and 129: 114 Appendix 18. Recording adverse
- Page 130 and 131: 116 Appendix 1level of benefit from
- Page 132 and 133: 118 Appendix 1• comparison of num
- Page 134 and 135: 120 Appendix 110.6 Sub-group analys
- Page 136 and 137:
122 Appendix 112. Data issues12.1 D
- Page 138 and 139:
124 Appendix 113. Organisation13.1
- Page 140 and 141:
126 Appendix 114. References[1] Ref
- Page 142 and 143:
128 Appendix 1Barnes RM, Gee L, Tay
- Page 144 and 145:
130 Appendix 1Appendix 1: Classifyi
- Page 146 and 147:
132 Appendix 1Liverpool re-treatmen
- Page 148 and 149:
134 Appendix 1Service StructurePlea
- Page 150 and 151:
136 Appendix 1Has your photographer
- Page 152 and 153:
138 Appendix 1There are many differ
- Page 154 and 155:
140 Appendix 1testing on vision tes
- Page 156 and 157:
142 Appendix 1something to do with
- Page 158 and 159:
144 Appendix 1College of Ophthalmol
- Page 160 and 161:
146 Appendix 1Appendix 5:Protocol f
- Page 162 and 163:
148 Appendix 16.2 Subjective Refrac
- Page 164 and 165:
150 Appendix 1• The patient shoul
- Page 166 and 167:
152 Appendix 1Appendix 6:Protocol f
- Page 168 and 169:
154 Appendix 18. It is customary to
- Page 170 and 171:
156 Appendix 1Early or Transit Phas
- Page 172 and 173:
158 Appendix 1
- Page 174 and 175:
160 Appendix 1---------------------
- Page 176 and 177:
162 Appendix 1---------------------
- Page 178 and 179:
164 Appendix 1---------------------
- Page 180 and 181:
166 Appendix 1Appendix 9:Site imple
- Page 182 and 183:
168 Appendix 1Option 2 - units inte
- Page 184 and 185:
170 Appendix 1Dos and Don’tsDosDo
- Page 186 and 187:
172 Appendix 1Resource use question
- Page 188 and 189:
174 Appendix 1Appendix 11: Recommen
- Page 190 and 191:
176 Appendix 1ADVERSE REACTION AND
- Page 192 and 193:
178 Appendix 1acute Trust; note tha
- Page 194 and 195:
180 Appendix 1Ophthalmologistrespon
- Page 197 and 198:
DOI: 10.3310/hta16060Health Technol
- Page 199 and 200:
DOI: 10.3310/hta16060Health Technol
- Page 201 and 202:
DOI: 10.3310/hta16060Health Technol
- Page 203:
DOI: 10.3310/hta16060Health Technol
- Page 206 and 207:
192 Appendix 3manufacturer’s pers
- Page 209 and 210:
DOI: 10.3310/hta16060Health Technol
- Page 211 and 212:
DOI: 10.3310/hta16060Health Technol
- Page 213 and 214:
DOI: 10.3310/hta16060Health Technol
- Page 216:
FeedbackThe HTA programme and the a