The clinical and cost-effectiveness of second-eye cataract surgery: a systematic review and economic evaluation

Introduction

The transparent ocular lens is enclosed in a membranous capsule and consists of two layers: a central ellipsoid, biconvex nucleus and a softer outer layer called the cortex. The lens contains epithelial cells that give rise to lens fibres throughout life, meaning that it becomes thicker and more compact with age and may lose optical clarity. Any loss of optical clarity in the lens is referred to as a cataract. Cataracts develop most often as a result of biological ageing, although there are numerous risk factors and some types of cataract are hereditary (see Cataract risk factors).

In the UK, one-quarter of the population will develop cataracts by the age of 75 years. 1 Cataracts lead to deterioration of vision, which may restrict daily activities and, in the elderly, independent living. Decreased visual function is associated with diminished quality of life,2 as well as a decrease in physical and mental health. 3 Poor vision related to cataracts is also a risk factor for falls and traffic accidents, which may lead to hospital admissions and limit independence. 4,5 The only effective treatment available to restore or maintain vision is the surgical removal of the cataract. 1 Cataract surgery is the most frequently conducted elective surgical procedure in the NHS, as well as in other Western countries. 6

Cataract types and classification

Cataracts are classified in a number of ways, primarily according to the part of the lens that is affected, and also according to their visual appearance and aetiology (e.g. some types of congenital cataract have a characteristic appearance). 7 The three most common types of age-related cataract are: nuclear sclerotic cataract, located in the lens nucleus; cortical cataract, located in the cortex; and posterior subcapsular cataract, located towards the back of the lens within the superficial posterior lens cortex. 7–9 Nuclear cataract is the most common type of age-related cataract. Posterior subcapsular cataract is more frequent among younger people.

Numerous clinical schemes have been developed for classifying the severity of the three main types of age-related cataract. Most of these schemes are based on slit lamp assessment of the eye (in which a narrow slit beam of light illuminates the eye) and comparison of the observations against standard cataract diagrams or photographs to determine a score. 10 The most popular classification schemes are the Oxford Clinical Cataract Classification and Grading System11 and the Lens Opacities Classification System III (LOCS III). 12 Alternative, simpler, schemes for recording cataract severity based on the area of the lens affected have also been used. 10

Prevalence and incidence of the condition

Recent data on the prevalence and incidence of cataracts in England and Wales are lacking. According to clinical experts, cataracts almost always develop bilaterally, i.e. patients with a cataract in one eye will already have signs of, or will develop, a cataract in their second eye. Estimates of the prevalence of people with cataracts in England have been made based on age- and sex-stratified regional population surveys,1 notably the North London Eye Study (1995–6), reported by Reidy and colleagues13 and Minassian and colleagues,14 and a study based on the Somerset and Avon survey of health (SASH) (1996–7), reported by Frost and colleagues,15 as well as other studies1 (described briefly below). These studies have produced markedly different estimates of prevalence, reflecting differences in their methodology, including different definitions of visual impairment and eligibility for cataract surgery.

The North London Eye Study randomly sampled people aged over 65 years from general practices in north London (1547/1840 people responded) and provides prevalence data specifically for visually impairing cataracts [i.e. Snellen visual acuity less than 6/12 that is attributable to a lens opacity, but not accounting for comorbidity or health-related quality of life (HRQoL)]. 13 The North London Eye Study found a prevalence of cataracts causing visual impairment (visual acuity in one or both eyes poorer than 6/12, attributable to lens opacity) of 30% [95% confidence interval (CI) 25.1% to 35.3%],13 and estimated the incidence of new cataract cases per annum in England and Wales to be around 224,000. 14 The prevalence of cataracts increased steadily with age from 16% in the 65–69 years age group, to 71% in people aged over 85 years. 13

The SASH-based study by Frost and colleagues15 randomly sampled 2783 people aged ≥ 55 years from 19 general practice registers in south-west England, from which data on 1078 people were available for analysis. Three definitions of eligibility for cataract were employed to reflect the likely range of eligibility criteria used in practice, and these included ocular comorbidity as well as HRQoL. 15 The estimated prevalence of cataract (in one or both eyes) requiring extraction was much lower than that from the North London Eye Study, ranging from 6 to 27 people per 1000 depending on the eligibility criteria. When adjusted according to the population profile of England (based on the 1991 census), the prevalent requirement for cataract surgery in England was estimated to range from 97,585 people (using the strictest eligibility criterion) to 384,000 people (using the least restrictive eligibility criterion), and was highest in patients aged over 74 years. 15

Recent data collected by the Royal National Institute of Blind People (RNIB)16 indicate that 269,636 people were treated for cataracts in England during 2012–13, which represents 0.48% of the population. This number of operations exceeds some of the cataract prevalence estimates from Frost and colleagues for England and Wales,15 indicating underestimation of the true prevalence if based on strict eligibility criteria for surgery. However, direct comparisons are difficult to make since the studies by Frost and colleagues15 and Reidy and colleagues13 were conducted over 15 years ago.

In a Medical Research Council (MRC) trial of assessment and management of older people in the community (1994–2001), nurses tested visual acuity in 14,403 people aged 75 years and older from 49 general practices in the UK. 17 Of 976 people with binocular visual impairment (binocular visual acuity < 6/18) excluding refractive error, 25% were classified as cataracts being the main or contributory cause of vision loss.

A study conducted in an urban population in England (the Newcastle 85+ cross-sectional study) based on family practice records of 839 patients aged 85 years and older in 2006–7 found that 36% had a previous history of cataract surgery in one or both eyes. 18 Surveys conducted both in the UK and the USA indicate that cataract prevalence is slightly higher in women,13,19 with 1.2 times as many women as men having cataracts in the North London Eye Study. 13 Prevalence also varies according to the type of cataract, with nuclear cataract being more common than posterior subcapsular cataract. 19

Assuming that it is reflective of the wider population case mix, the Cataract National Dataset, which reflects 55,567 cataract operations conducted in 12 NHS trusts during November 2001 to July 2006, would suggest that the ‘typical’ population in England presenting for cataract surgery in either eye would be 62% female and of a mean age of 75.4 years. 20 Pre-operative visual acuity was 6/12 or better in 55.3% of those patients surveyed who were attending second-eye cataract surgery. However, it should be borne in mind that visual acuity thresholds may have declined further since these data were collected. 20

Cataract risk factors

Although cataracts are more common in older people, the risk of developing cataracts may be associated with a wide range of factors. These include genetic factors,21 diet,22 metabolic and nutritional disorders (e.g. diabetes mellitus, thyroid disorders), and treatment of ocular diseases (e.g. glaucoma) with steroids or vitreoretinal surgery with heavy gases or oil in the eye. Extrinsic risk factors include infections, smoking, alcohol and the use of systemic corticosteroids. 23 Exposure to radiation also increases the risk of developing cataracts, as the ocular lens is particularly sensitive to the effects of ionising radiation. Groups exposed to radiation that are at increased risk of developing cataracts include patients who have received radiotherapy, clinicians who frequently use X-ray procedures and airline pilots exposed to ultraviolet wavelengths. 24,25 As the prevalence data referred to above suggest, women appear to have a slightly higher risk of developing cataracts than men. 13,17,19,23

Symptoms and diagnosis

Generally, age-related cataracts are progressive8 although during the early stages of cataract development there may be minimal changes to an individual’s visual function. As cataracts develop further, they can affect vision in a number of ways, including increasing myopia (nearsightedness) and astigmatism, inducing monocular diplopia (double vision), reducing light transmission and/or changing colour perception. Visual loss from cataracts is principally as a result of the decreased transmission of light through the lens. Cataracts also increase scatter of intraocular light, which reduces the contrast of the retinal image, with the visual decrement being greatest in glare or bright light conditions. 10 During the later stages of opacity, vision may be limited to light and dark distinction. Cataract symptoms include blurred or reduced vision, and problems associated with glare or low-contrast conditions, ultimately leading to blindness if untreated, but symptoms do depend on the location and size of the cataract, and whether cataracts are present in one or both eyes. For example, glare may be a particular problem for people with posterior subcapsular cataracts, although they may see well in dim illumination, whereas people with nuclear cataracts may experience myopic refractive shift (improved near vision but worse distance vision), difficulty with night driving and loss of colour discrimination ability. 7

There is no single test to assess the effect of a cataract on a patient. Diagnosis usually involves examination of the eye by an ophthalmologist or optometrist using a slit lamp and assessment of one or more aspects of visual function, including visual acuity, contrast sensitivity and stereopsis (described in more detail in Clinical vision outcome measures). Clinical decisions concerning the need for cataract surgery also take into account the extent of symptoms and the impact of a cataract on the patient’s well-being and quality of life (see Current service provision).

Measurement of the condition

Approaches for measuring the impact of cataracts on patients can be divided into those that assess impacts on clinical vision and those that assess impacts on the way patients feel, for example how cataracts affect patients’ symptoms and quality of life. The main clinical vision measures that have been used for assessing effects of cataracts, both in clinical settings and as outcomes in clinical trials, are visual acuity, contrast sensitivity and stereopsis (other vision outcomes such as glare disability may be important to patients, but are less commonly reported in clinical trials of cataract surgery). A wide variety of validated patient-reported outcomes relating to symptoms and quality of life have been used in clinical trials but, unlike measures of clinical vision, they are not routinely employed in clinical decision-making practice for cataract surgery. The three main clinical visual outcomes and some of the most frequently used patient-reported outcomes that are used for assessing impacts of cataracts in clinical trials are described below.

Clinical vision outcome measures

For the outcome measures described below, assessments may be made with one eye covered (monocular assessment) or with both eyes open (binocular assessment). In addition, assessments may be made under specific conditions, for example to test for vision unaided or with spectacles, or to test for near or distance vision.

Patient-reported outcome measures

A variety of instruments have been employed in research studies for assessing the impacts of cataract surgery on patients’ symptoms, functional ability, well-being and HRQoL. 1 These can be divided into generic instruments which have been designed to apply across a range of different health conditions [e.g. the European Quality of Life-5 Dimensions (EQ-5D) health survey and Short Form questionnaire-36 item health survey instrument (SF-36)] and instruments that focus specifically on vision-related conditions [e.g. the Visual Function Index (VF-14)]. Examples of vision-specific instruments are also provided by Khadka and colleagues47 and McAlinden and colleagues. 48

Impact on patients and the NHS

Without surgery, most types of cataract will progress with time, leading to loss of visual function, although the rapidity and severity of visual decline depends on the type and severity of cataract present as well as any ocular comorbidities (see Surgical and post-operative complications). Untreated cataracts may ultimately lead to blindness. Benefits of cataract surgery include improved visual acuity, with 85–90% of people having 6/12 best corrected vision as measured on a Snellen chart (i.e. meeting the driving requirements in the UK); improved clarity of vision; and improved colour vision. 9

Cataracts can affect patients directly by impairing their vision (e.g. visual acuity, contrast sensitivity and/or stereopsis) and indirectly by the deterioration in functional abilities and quality of life that can result from the visual impairment. In older people, deterioration of vision as a result of cataracts can have a range of consequences. People may have difficulty performing daily tasks, reading, using computers, or participating in games or social activities. Elderly women with cataracts are at an increased risk of having falls and fractures. 4

An issue of particular concern among drivers is that cataracts can affect their ability to drive safely. 16 Studies comparing populations of cataract patients who have or have not received cataract surgery suggest that cataracts contribute to self-reported driving difficulty49 and increase the frequency of motor vehicle collisions, as well as the financial costs associated with them. 50 It has been suggested that providing cataract surgery earlier, rather than after significant visual problems have become apparent, could reduce traffic accidents among older drivers. 5 Vision-related driving requirements are usually based only on visual acuity,51 although drivers who pass a visual acuity test may have poor contrast sensitivity or other visual defects which, although not tested, can impair their driving ability. 52,53 In the UK, people are required have a Snellen distance visual acuity of at least 6/12 (with spectacles if appropriate) in order to be allowed to hold a driving licence. 51 In addition, since 2001 the legal requirement for standards of vision for driving has been based on being able to read a car number plate from 20 metres. This may overestimate the driver’s functional vision in real-life motoring conditions, as many drivers with early cataracts experience problems with night vision. According to clinical experts consulted during this project, in cases where cataract surgery causes enough improvement in visual function for drivers to retain their eligibility to drive, this may impact positively on their quality of life. However, there do not appear to be any data on how many patients this would apply to, nor on the extent of the impact on quality of life.

Cataract surgery is the most common elective surgical procedure performed in the NHS,1 and cataracts place a direct financial burden on the NHS in terms of the cost of treatment. However, performing cataract surgery may lead to savings in health-care costs elsewhere, for example those associated with falls and traffic accidents, although it is uncertain how important second-eye cataract surgery would be compared with first-eye surgery in this respect.

According to a report by the RNIB published in 2013,16 there were 340,809 cataract operations in England during 2012–13 (up to 28 February 2013). The most recent data available on the proportions of first- and second-eye cataract surgery in England are from 2011–12 in the RNIB report,16 and indicate that cases of second-eye cataract surgery that were conducted within 1 year of the first-eye surgery accounted for 27% of all cataract operations. The National Schedule of Reference Costs54 puts the total cost of cataract surgery for 2011–12 at £240M. Assuming that 27% of cataract operations are on the second eye,16 an estimate of the cost per annum of second-eye surgery alone would be £64.8M. The burden of cataracts to the NHS will increase as the population ages and larger numbers of older people seek treatment. However, figures reported by the RNIB suggest that the frequency of second-eye cataract operations has decreased in England from 96,336 in 2009–10 to 91,959 in 2011–12. 16

Relevant national guidelines

Current guidance on referral criteria for cataract surgery is given in the Department of Health report Action on Cataracts, which was published in 2000. 62 The report led to a number of improvements in NHS cataract care pathways during the 2000s, including enabling direct referral by optometrists, simplifying the number of consultations required, increasing the throughput of patients and reducing waiting times, and emphasising the need for care to be patient centred to prevent situations where people experience a reduction in their quality of life owing to treatable cataracts. 59,62 The most relevant current national guidelines are the Royal College of Ophthalmologists Cataract Surgery Guidelines (2010) (next due for review in 2015) which state that it is clinically and economically appropriate for second-eye surgery to be offered to those patients who want it. A 1b recommendation is given (a recommendation based on at least one randomised trial). 1 Similarly, the Scottish Health Technologies Group issued a statement in September 2012 advising that there is randomised controlled trial (RCT) evidence to support second-eye surgery and evidence from cost–utility analysis to demonstrate lifetime cost-effectiveness. 63 In England and Wales, the National Institute for Health and Care Excellence (NICE) has not previously appraised second-eye cataract surgery. It is thought that such an appraisal is unlikely to be conducted by NICE until 2018, although the RNIB and Royal College of Ophthalmologists have requested this to be prioritised. 16

Surgical and post-operative complications

The most frequent adverse consequence of cataract surgery is a clouding of the back of the lens capsule, known as posterior capsular opacification (PCO), which can lead to deterioration in vision months or years after the initial surgery. For the purposes of this report, we refer to PCO as a complication of cataract surgery, although strictly speaking PCO could be considered an expected consequence of phacoemulsification in a proportion of patients since phacoemulsification leaves the lens capsule intact but prone to opacification after cataract extraction. Clinical experts estimated that 20–30% of patients who have cataract surgery will experience PCO. Except in rare cases, PCO is easily treated with a procedure known as neodymium-doped yttrium–aluminium–garnet (Nd:YAG) laser lcapsulotomy in which a laser with a Nd:YAG crystal is used to cut a hole in the opacity of the lens capsule to allow light to pass through. 64 Laser capsulotomy is a painless outpatient surgical procedure that does not require anaesthesia. A single laser capsulotomy can provide lifelong recovery of good vision, although complications may occur in a minority of patients. 64

Serious surgical complications associated with cataract surgery (which may be sight-threatening, or increase the risk of sight-threatening sequelae) are posterior capsule rupture (PCR) during surgery65 and, post operatively, endophthalmitis,66 retinal detachment,67,68 cystoid macular oedema69 and retained lens fragments. 70,71 These complications are relatively uncommon, affecting fewer than 2% of cataract operations. 1,20,68,72–74 However, the occurrence of PCR during surgery increases the risk of cystoid macular oedema, endophthalmitis and retinal detachment. 65

Other complications that may arise from cataract surgery9 but which are usually not serious, except in rare cases, are uveitis, transient increase in intraocular pressure, bruising of the eye or eyelids after surgery which usually resolves without problems, refractive surprise (an unexpected need for a particular glasses prescription) and corneal decompensation (corneal oedema).

Identification of important subgroups

Important subgroups that may affect cataract surgery outcomes include patients’ age, pre-operative ocular comorbidities and the experience of surgeons conducting the procedure. Data from the UK Cataract National Dataset indicate that pre-operative risk factors for PCR and/or vitreous loss during cataract surgery include (among others) increasing age, male gender, glaucoma, diabetic retinopathy and trainee surgeons performing operations. 75 Other pre- and peri-operative adverse risk factors for post-surgical monocular vision loss are age-related macular degeneration, amblyopia, corneal pathology, previous vitrectomy and posterior capsule rupture during surgery. 76 The UK Cataract National Dataset found that, during 2001–6, 28.5% of people presenting for cataract surgery (one or both eyes) had one or more ocular comorbidities. Post-operative visual acuity of 6/12 or better was achieved in 97% of patients who had no pre-operative comorbidities and 79.9% of patients with one or more ocular comorbidities (51% and 30%, respectively, achieved visual acuity 6/6 or better). 20

Clinical vision outcomes

None of the RCTs reported the severity, grade or type of cataract that their patients had. However, Foss and colleagues83 stated that cataracts in their study were less severe than those in the earlier trial by Laidaw and colleagues. 84

Patient-reported outcomes

There was a large variation in patient-reported outcomes between the studies, with few outcome measures common to more than one RCT. Although general health status was reported by all three studies, Laidlaw and colleagues84 utilised the SF-36, Foss and colleagues83 reported using the EQ-5D, and Castells and colleagues85 the Short Form questionnaire-12 item health survey instrument (SF-12). Both Foss and colleagues83 and Castells and colleagues85 assessed patient-reported visual disability as an outcome using the VF-14 instrument.

Falls and fractures

Only Foss and colleagues83 reported falling (in a female population) as an outcome (Table 10). A fall was defined as unintentionally coming to rest on the ground or at a lower level with or without loss of consciousness. The proportion of women experiencing falling during the 12-month follow-up period was higher in the expedited surgery group (40%) than in the routine surgery group (34%). However, the authors did not report a statistical comparison between the treatment groups. The hazard ratio for women experiencing a first fall was 1.06 at 6 months follow-up, but not statistically significant, equating to a non-significantly higher rate of first falls in those who had received second-eye surgery (22%) than in women awaiting second-eye surgery (16%), whereas the number of women experiencing more than one fall was the same in both groups (18%). The rate of falling per 1000 patient-days was also non-significant. Fractures were experienced in 4% of women who had received second-eye surgery and in 3% of women awaiting second-eye surgery.

Surgical complications

Two of the studies reported surgical complications. 83,84 It should be noted that extracapsular cataract extraction, which was the method employed by Laidlaw and colleagues,84 has historically been associated with more problematic complications than the phacoemulsification approach to cataract surgery employed in the more recent study by Foss and colleagues. 83

Laidlaw and colleagues84 reported six types of surgical complication which were either sight-threatening or necessitated further surgery. In total, 11 complications occurred in nine patients (9%) in the expedited surgery group (Table 11). The most common complications were cystoid macular oedema (three events) and iris prolapse (three events).

Foss and colleagues,83 reported seven types of surgical complications, although two (section stitched and iris hooks) are surgical procedures rather than complications and we have not extracted these. The most frequent complication in the expedited surgery group was PCO at 6 months (10%), with Nd:YAG laser capsulotomy (to treat PCO) being performed at any time during the study period in 9% of the patients (see Table 11).

A limitation of these data is that it is not clear how complete the reporting of complications was, since follow-up differed between the RCTs and different complications were listed in each RCT. Foss and colleagues83 had a 12-month follow-up for most clinical outcomes but, as noted above, assessed only incidence of PCO 6 months after surgery, whereas Laidlaw and colleagues84 had a 6-month follow-up. The frequencies of complications such as PCO and retinal detachment, which may occur more than a year after cataract surgery, might therefore have been underestimated in both RCTs. It is unclear whether Laidlaw and colleagues84 had no cases of PCO up to 6 months after surgery or whether they did not record this complication. The RCT of Laidlaw and colleagues84 was conducted before phacoemulsification was introduced, and the complications they reported with extracapsular cataract surgery might not be reflective of current practice, as the safety of cataract surgery has improved since the introduction of phacoemulsification. 88 Among the complications reported in these RCTs, the most serious are retinal detachment and those involving vitreous loss or vitrectomy.

Castells and colleagues85 reported the number of post-randomisation exclusions that were due to complications of first-eye surgery. There were two exclusions from the second-eye surgery group and one from the single-eye surgery group.

Adverse events

Other than surgical complications, the only other adverse event that was reported in the RCTs was mortality. A patient flow chart presented by Laidlaw and colleagues84 suggests that no mortality occurred during their RCT. Foss and colleagues83 reported that three patients died during their RCT: one in the expedited surgery group and two in the routine practice group. Castells and colleagues85 also reported that three patients died during their RCT, of whom one was in the single-eye surgery group and two of whom were in the second-eye surgery group. Causes of mortality were not reported in any of the studies.

Ongoing studies

No ongoing studies were identified up to 10 July 2013.

Relevant systematic reviews

No relevant systematic reviews were identified in the study protocol (see Appendix 1) or during bibliographic searches conducted up to July 2013. However, after data synthesis had been completed, we identified a systematic review, published in October 2013, of the benefits of second-eye cataract surgery in the elderly. 89 Ishikawa and colleagues’89 inclusion criteria did not limit studies to RCTs, although their searches were limited to English-language publications. They included 10 primary studies, of which three were the RCTs that we report above,83–85 five were cohort studies, and two were before-and-after studies (no other RCTs were identified). One of the cohort studies they included, by Räsänen and colleagues,90 also investigated cost-effectiveness and HRQoL and is included in our systematic reviews of cost-effectiveness and HRQoL (see Systematic review of cost-effectiveness evidence and Systematic review of health related quality-of-life studies).

Ishikawa and colleagues89 concluded that second-eye surgery led to improvement in visual acuity, contrast sensitivity, stereopsis, and self-reported visual functioning, supported by moderate-quality evidence. They also concluded that there was improvement in HRQoL based on mixed-quality evidence, but no improvement in falls reduction, based on limited-quality evidence. Interpretation of these findings is not straightforward, however, since the assessment of evidence quality was based on ‘best evidence synthesis guidelines’ which assessed only whether or not an improvement in outcomes occurred and did not consider the magnitude and variance of any observed improvements. Moreover, no quantitative results were presented and it was not specified whether the reported improvements in outcomes were based on clinical importance, statistical significance or other criteria. 89

Study characteristics

Three parallel-group RCTs conducted from 1994 to 2006, each randomising 208–296 elderly participants (mean age 71.1–79.9 years), met the selection criteria and were included in the systematic review of clinical effectiveness. 83–85 The RCTs varied in quality, with all three being deemed at high risk of performance bias (patients could not be masked to the treatment group).

One RCT was conducted before phacoemulsification was introduced in the UK84 and one RCT included only women;83 the RCTs also differed in the timing of their baseline assessments. Owing to this methodological heterogeneity, quantitative pooling of outcomes in a meta-analysis was deemed inappropriate. Instead we synthesised outcomes narratively.

Effects on vision

All three RCTs reported measures of visual acuity, contrast sensitivity and stereopsis, but they did not report any other measures of clinical vision (e.g. glare disability). The three RCTs all indicated statistically significant slight improvements in binocular visual acuity (ranging from 0.04 to 0.063 log-MAR) after second-eye surgery compared with surgery in one eye alone. However, these differences (less than 1 Snellen line equivalent) would not be considered clinically important. Changes in monocular visual acuity in the second eye after surgery were reported in two RCTs and were relatively large (0.46 log-MAR83 and 0.75 log-MAR84), although neither RCT stated whether or not the changes were statistically significant.

Binocular contrast sensitivity was higher after second-eye surgery in all three RCTs, with the difference statistically significant in two RCTs. 83,84 However, the differences were clinically small, equivalent to 0.09 dB (less than 0.05 log-units)83 or no more than four letters on a Pelli–Robson chart. 84

Baseline stereopsis differed among the RCTs, being notably poor in the RCT by Laidlaw and colleagues,84 with 61–71% of patients having no functional stereopsis (3000 seconds of arc or worse). In contrast, in the RCT by Foss and colleagues83 65–76% of patients were in the best stereopsis class (150 seconds of arc) at baseline. Castells and colleagues85 did not collect baseline data immediately before second-eye surgery, but their data collected after first-eye surgery (a proxy for baseline data in the second-eye surgery group) indicate moderate stereopsis (mean approximately 230 seconds of arc). Stereopsis significantly improved after second-eye surgery in all three RCTs, although different ways of presenting and analysing these data were used in each RCT. These improvements would be of clinical benefit to patients’ vision and were largest in the RCT by Laidlaw and colleagues84 whose patients had the worst baseline stereopsis.

Effects on patient-reported outcomes

Three RCTs measured patient-reported outcomes for general HRQoL, but each used a different preference-based instrument (SF-12, SF-36 or EQ-5D). 83–85 Of these, the mental health component summary of the SF-12 indicated a statistically significant and clinically relevant improvement after second-eye surgery. 85 However, differences in the other measures between first- and second-eye surgery were not statistically significant (EQ-5D, SF-36)83,84 or the statistical significance was not reported (physical health component summary of the SF-12). 85 One RCT, which focused specifically on elderly women,83 employed generic instruments to assess specific domains of HRQoL and psychological distress. No statistically significant differences were found between first- and second-eye cataract surgery for activities of daily living, anxiety or depression. Statistically significant differences between the study groups were found for falls efficacy and handicap, although the differences in scores (3.6 on the Falls Efficacy Scale and 4.4 on the London Handicap Scale) were modest in comparison with already high baseline scores. The clinical importance of these findings is unclear, as the MCID has not been established for these patient-reported outcome measures and the authors of the RCT did not discuss clinical interpretation. 83

The only validated vision-specific instrument used was the VF-14, which assesses the impact on 14 vision-dependent activities. Two RCTs that employed this instrument recorded significantly higher (better) scores in the group who had received second-eye surgery, with mean differences of 7.5 points83 and 8.24 points. 85 However, these differences are smaller than the proposed MCD and MCID for the VF-14,87 suggesting that they did not reach clinical significance.

In summary, the majority of patient-reported outcome measures employed by the three RCTs did not demonstrate convincing effects of second-eye cataract surgery on patients’ HRQoL. This might reflect relatively good visual acuity and HRQoL that many patients had before second-eye surgery (i.e. ceiling effects of these outcomes) and/or limitations of the assessment instruments that were employed (see Chapter 7).

Effects on falls and fractures

Falls and fractures were assessed only in one RCT. 83 Compared with first-eye cataract surgery, second-eye surgery did not reduce the number of falls or fractures significantly.

Surgical complications

Two RCTs reported surgical complications, although one84 was conducted before phacoemulsification was introduced in the UK. In the more recent RCT, more relevant to current practice,83 the reported complications were typical of those associated with phacoemulsification, the most frequent being PCO (10% at 6 months).

Critical appraisal of the studies

The cost-effectiveness studies were assessed against a critical appraisal checklist (Table 13). This checklist assessed the quality of the studies and their generalisability to the UK and was adapted by the review authors from checklists by Philips and colleagues,80 Drummond and colleagues81 and the NICE reference case requirements. 95

Two of the cost-effectiveness studies by Räsänen and colleagues90 and Sach and colleagues,91 were also included in our systematic review of HRQoL (see Systematic review of health-related quality-of-life studies for a more detailed consideration of their HRQoL results).

All studies clearly defined the study question and explained the competing alternative. Two studies used a relevant comparator,91,92 whereas the other used a comparator that would be unlikely to be used in the UK (a hypothetical no-treatment group). 90 Each study clearly stated the patient group of interest and the study perspective. One of the studies was conducted in the UK,91 and the other studies were conducted in the USA92 and Finland,90 and it is unclear how generalisable these studies are to the UK NHS. The study type and methodology are appropriate for the decision problem in this report (see Chapter 2).

Reported descriptions of the parameters and assumptions varied between studies. Räsänen and colleagues90 provided limited information on assumptions and costs in their analysis, whereas the other two studies91,92 provided more complete information on their inputs and assumptions.

Busbee and colleagues92 based the effectiveness of the intervention on a single study, rather than a systematic review. They also included HRQoL, using standardised and generic instruments and reported the health benefits in terms of QALYs. Two studies91,92 discounted costs and benefits. The other study90 only discounted costs, as all costs were incurred in the first 6 months. All three studies assessed uncertainty through sensitivity analyses. Busbee and colleagues92 did not give any details of whether or not their model was validated.

Overall, the studies were found to vary in methodology quality. The studies by Busbee and colleagues92 and Sach and colleagues91 were of reasonable methodological quality, whereas Räsänen and colleagues’ study90 had limitations in its reporting and assumptions. The study by Sach and colleagues was of most relevance to the UK NHS91 (see Table 13).

Busbee and colleagues

Räsänen and colleagues

Sach and colleagues

Comparison of studies

Cost-effectiveness results varied across the studies. Busbee and colleagues92 reported an ICER of US$2495 per QALY,92 and Sach and colleagues91 reported an ICER of £44,263. However, in the latter study the ICER was reduced to £17,299 per QALY when a lifetime horizon was used. Räsänen and colleagues90 did not report an ICER, as second-eye cataract surgery was associated with negative QALYs.

The cost-effectiveness results were driven by the changes in utility in the studies. Each study used a different HRQoL instrument (EQ-5D, TTO or 15D). Post-surgical changes in HRQoL varied between the studies, with a utility gain of 0.109 in one study92 and a utility loss of 0.01 in two studies. 90,91 Räsänen and colleagues90 suggested that the reason for the lack of impact on HRQoL could be that two-thirds of the patients reported only minimal pre-operative subjective seeing problems despite objective evidence of poor visual acuity in the surgical eye, and one-third of patients had a secondary ophthalmic diagnosis which might have reduced the benefit of cataract surgery.

There was variation between the studies in the degree of pre-surgical visual impairment in the second eye. For Busbee and colleagues92 and group C of Räsänen and colleagues (second-eye patients),90 the mean pre-surgical visual acuity was 0.24 decimal (equivalent to around 6/24 metres), whereas in the Sach and colleagues’ study91 pre-surgical visual acuity was 0.09 decimal (or 6/7 metres).

The studies differed in their approaches to measuring the costs of second-eye surgery. Busbee and colleagues’92 report was the only study to assess the costs associated with treating post-operative complications, although the other studies collected hospital costs after surgery. Sach and colleagues91 conducted the only study to include the post-surgery costs associated with social services, such as residential care costs. Two of the studies extrapolated the short-term costs to a lifetime horizon by assuming future costs to be similar to those seen during the trial period, whereas Räsänen and colleagues90 used only the costs of the cataract procedure and hospitalisation and did not consider further costs.

In summary, there was variability across the three studies in terms of modelling approaches, patient characteristics, assumptions and parameter values, and this led to substantial differences in the estimated cost-effectiveness of the second-eye surgery. Therefore, no existing model was considered appropriate for this HTA and a de novo model was developed (see Methods for economic analysis).

Characteristics of the included studies

Half of the studies (Table 14) were conducted in Europe,40,90,91,101,106 of which two were from the UK. 40,91 One of the UK studies, by Sach and colleagues,91 was a cost-effectiveness study conducted alongside a RCT and was included in our systematic review of cost-effectiveness (see Systematic review of cost-effectiveness evidence), with the RCT itself (Foss and colleagues83) included in our systematic review of clinical effectiveness (see Chapter 4). Of the remaining studies, two were from the USA,102,104 one from Japan94,103 and one from Australia. 105

Study designs included RCTs with economic evaluations,91,101,104 and RCTs with HRQoL evaluations,40 with the remainder of the studies being observational. The observational designs comprised single cohorts of patients;86,103,106 single cohorts of patients with subgroups based on whether patients received first-eye, second-eye or bilateral surgery;90,94 and a case–control study. 105

Mean ages, in years, in the studies were generally in the early 70s, but ranged from 69 to 81 years. The majority of the patient populations studied were mixed sex, with the exception of two studies in which all patients were female. 40,91 In the majority of the studies in which baseline clinical vision status was reported, the patients could be classed as having poor visual acuity. The mean visual acuity (generally in the eye to be operated on) was most often below Snellen 6/12, indicating visually impairing cataracts. 107 One study was an exception,91 with patients described as having minimal visual dysfunction in the eye to be operated on (86% had a baseline visual acuity of 6/12 or better).

A range of HRQoL instruments were used by the studies, the most common being the EQ-5D (seven studies). 40,86,91,94,101,105,106 Four studies used the TTO method with patients to elicit utility values,94,101,103,105 three studies used the Health Utilities Index 3 (HUI-3)86,94,103 and two studies asked patients to rate their HRQoL using a visual analogue scale (VAS) [in one case the EuroQol-visual analogue scales (EQ-VAS);106 in the other study101 the VAS instrument was not specified].

The studies were classified into three groups based on the comparisons it is possible to make (Table 15): studies in which a comparison between first- and second-eye cataract surgery is possible;90,91,94,101 non-second-eye cataract surgery comparisons (including, but not restricted to, studies assessing only first-eye surgery);40,86,103,104 and studies assessing the impact of cataract surgery complications on HRQoL. 105,106

The following instruments were used in one study each: HUI-286, Short Form questionnaire-6 Dimensions (SF-6D),86 the Quality of Well-Being – Self-Administered (QWB-SA) questionnaire86 and the 15D. 90 It was common for studies to use more than one instrument, for purposes of comparison. For example, Feeny and colleagues86 compared agreement among five generic preference-based measures of HRQoL in a cohort of patients undergoing cataract surgery. Some of the studies used other instruments to assess patient symptoms and HRQoL but data are not presented for them here (in line with our inclusion criteria) as they were vision-specific rather than generic health instruments and/or they were generic health instruments that did not yield health utilities. These included the VF-14,40 the Visual Function Questionnaire-25 item (VFQ-25),105 the Activities of Daily Vision Scale,104 the SF-12,104 Charlson Comorbidity Index104 and the London Handicap Scale40 (see Table 15).

Results of the included studies

This section focuses on the results of the four studies which provided second-eye cataract surgery comparisons, as these are most relevant to the decision problem of this report. The results of these and the other six studies are summarised in Table 15.

Dolders and colleagues101 reported a cost–utility analysis comparing monofocal and multifocal intraocular lenses (IOLs) in cataract surgery in the Netherlands. Patients were randomised to receive surgery with either a monofocal or a multifocal IOL (n = 143 analysed). The patients were interviewed to assess HRQoL and all other outcomes at three time points: T1 (1–2 weeks before first-eye surgery), T2 (3 months after first-eye surgery) and T3 (3 months post operatively after second-eye surgery). Although not an intention of the authors, it is possible to examine changes in utility values from T2 to T3 in both randomised arms, analogous to assessing the effect of second-eye surgery. Utility values were elicited through use of VAS, TTO and SG (the EQ-5D was used, but utility values were not reported in the study publication). Although there was a general increase in utility from baseline (T1) to post-first-eye surgery (T2), utility values declined (in both trial arms) between first- (T2) and second-eye surgery (T3). This decline was common to all three utility instruments used.

Hiratsuka and colleagues94 assessed HRQoL using the TTO, the EQ-5D and the HUI-3 prior to, and 3 months after, cataract surgery. Patients were recruited from Japanese hospital ophthalmological clinics and private surgical clinics, with the aim of reflecting a real-world setting. Patients were classified using three subgroups: first-eye surgery (n = 157), second-eye surgery (n = 60) and bilateral eye surgery (n = 312). For second-eye surgery patients, increases in utility values ranged from 0.08 (HUI-3) to 0.24 (TTO) and were statistically significant for all instruments. Generally, utility gains were slightly higher for second-eye cataract surgery patients compared with first-eye patients (except on the HUI-3), and utility gains for bilateral eye surgery were slightly higher compared with first- and second-eye cataract surgery (except on the EQ-5D) (see Table 15). However, differences between these subgroups were not tested statistically. The authors suggest that the TTO method, when used to measure ophthalmic treatment, is more sensitive to small changes in utility than generic survey questionnaires such as the EQ-5D and HUI-3. However, they note that even the relatively smaller utility gains for EQ-5D and HUI-3 were statistically significant.

The study by Räsänen and colleagues90 was similar in design to that of Hiratsuka and colleagues,94 comprising three subgroups of Finnish patients: first-eye surgery (group A, n = 87); bilateral eye surgery (group B, n = 73), and second-eye surgery (group C, n = 59). HRQoL was measured using the 15D instrument, at baseline and at the 6-month follow-up. There was no change in utility for first-eye surgery patients, a slight increase of 0.03 for bilateral eye surgery patients, and a reduction of 0.01 for second-eye surgery patients. The only statistically significant result was for the bilateral surgery group. The authors propose a number of explanations for lack of improvement in HRQoL, including the fact that two-thirds of patients reported only minimal pre-operative subjective seeing problems, despite objective evidence of poor visual acuity in the surgical eye (though at baseline there was poor correlation between 15D score and visual acuity in the surgical eye); the real-world setting of a university clinic and its mixed sample; that one-third of patients had a secondary ophthalmic diagnosis (which might reduce the benefit of surgery); and potential insensitivity of the 15D to measure changes in HRQoL (it includes only one question relating to sight).

Sach and colleagues,91 as reported earlier, conducted a cost-effectiveness study alongside a RCT in the UK comparing expedited second-eye surgery (n = 116, target within 1 month) with a waiting list control group (n = 113, described as routine surgery, target within 13 months). All patients were female and described as having minimal visual dysfunction in the eye to be operated on (approximately 86% of participants had baseline visual acuity of 6/12). There was a reduction in utility in both arms following surgery: −0.01 and −0.03 for expedited second-eye and unoperated second-eye surgery patients, respectively (representing a 0.02 gain for second-eye cataract surgery patients). The difference between arms was reported to be not statistically significant. The authors discuss the potential limitations of using EQ-5D to assess the benefits of cataract surgery, noting that it does not incorporate sensory function in its descriptive system and, therefore, may lack precision and responsiveness to changes in HRQoL.

Discussion

There was variation in the utility estimates across the studies included in the review. Räsänen and colleagues90 reported a decrease in utility of 0.01 following second-eye surgery compared with no change for patients undergoing first-eye surgery. Sach and colleagues91 also reported a decrease of 0.01 for second-eye surgery compared with a slightly bigger decrease of 0.03 for patients not undergoing second-eye surgery, representing a net 0.02 utility gain. Dolders and colleagues101 reported decreases in utility of 0.01 to 0.08 from completion of first-eye surgery to completion of second-eye surgery, with estimates varying according to which HRQoL instrument was used. In their study, for first-eye cataract surgery there was an increase in utility for some instruments, no change for other instruments and for some instruments there was a slight decrease.

Hiratsuka and colleagues94 reported statistically significant increases in utility following second-eye surgery, with the magnitude of the increase varying according to which instrument was used. The EQ-5D and the HUI-3 both yielded gains of 0.09, whereas the TTO yielded a larger value of 0.24. The TTO estimates were derived directly from preferences valued by cataract surgery patients themselves, while the EQ-5D and HUI-3, though completed by patients, use preferences valued by samples of the general public. The EQ-5D does not include a sensory function and may not always be sensitive to improvements in clinical vision following cataract surgery. 90,91,108 However, the HUI-3 does contain a visual utility subscale and may be more responsive to change. For example, the study by Feeny and colleagues86 (presumed to include first-eye cataract surgery patients, though not explicitly stated) compared a number of generic HRQoL instruments and found that the largest increases in utility were for the HUI-3 and for subscales for sensation (HUI-2 sensation) and vision (HUI-3 vision) (see Table 14). Feeny and colleagues86 suggest that a change of 0.03 or more in the utility score for preference-based measures is interpreted as a clinically important change for cataract surgery. They also propose that for the single-attribute utility scores for HUI-2 sensation (which includes vision) and HUI-3 vision, the guideline for a clinically important difference is 0.05. The utility estimates reported by Hiratsuka and colleagues94 therefore exceed these thresholds for clinically important differences and may be suitable for inclusion in cost-effectiveness analysis.

Of the four studies, those by Sach and colleagues91 and Hiratsuka and colleagues94 are considered to be the most appropriate for informing the assessment of cost-effectiveness of second-eye cataract surgery in this HTA, based on methodological quality. Each study has advantages and disadvantages. To summarise, Sach and colleagues91 benefited from a RCT design directly comparing patients receiving second-eye surgery with patients on a waiting list. The study was conducted in the UK and the utility estimates could therefore be considered more relevant to the NHS. However, the patient population was exclusively female, and had mild visual impairment, so in this respect is less typical than the UK patient population where visual impairment is greater. 20 Furthermore, the authors report patients’ binocular rather than monocular visual acuity at baseline, prohibiting an assessment of the degree of visual impairment in the eye to be operated on (the second eye). The study also assessed HRQoL using the EQ-5D which, as discussed above, does not include a sensory function. Similarly, Hiratsuka and colleagues94 employed the EQ-5D but also used the TTO method and the HUI-3, which has a visual utility subscale and is therefore potentially more sensitive to improvements in vision following cataract surgery. Hiratsuka and colleagues94 also reported baseline monocular visual acuity for both eyes, allowing an assessment of the degree of visual impairment in the eye to be operated on. However, the study was observational in design, and patients undergoing second-eye cataract surgery were a relatively small subgroup of a wider group of patients having first-eye or bilateral cataract surgery. Therefore, reported visual acuity is not specific to second-eye cataract surgery patients. Furthermore, this was a Japanese study and applicability to the UK patient population is questionable. Given that expert clinical advice sought in the production of this report suggests that use of a HRQoL instrument that is sensitive to changes in clinical vision is a key consideration, on balance Hiratsuka and colleagues94 would be the most appropriate study for informing the assessment of cost-effectiveness in this report, as will be described in the next section.

The remaining two studies which reported changes in HRQoL following first- and second-eye surgery (see Table 14) were not considered appropriate for informing the base case of our cost-effectiveness analysis. Räsänen and colleagues90 used a HRQoL utility measure, 15D, not commonly used outside Scandinavia. The utility loss for second-eye surgery was small and there was no statistical difference between the utility at baseline and at 6 months. Furthermore, it is unclear whether or not this utility loss was due to other factors such as general health utility decline for an old population, as the study was for a single cohort. Dolders and colleagues101 employed several preference-based instruments (see Table 14) which provide relevant utility estimates but it is unclear whether the study could reliably detect changes in utilities since no statistical tests of differences in relation to first- and second-eye surgery were reported. This may reflect that the aim of the study was to compare two different types of IOL, not to compare first- and second-eye surgery. Nevertheless, the study by Dolders and colleagues101 does provide information on utility changes in relation to first- and second-eye surgery within each of the IOL comparison groups (albeit of uncertain reliability) and the results suggest that second-eye surgery was consistently associated with a decline in utility values compared with first-eye surgery. For this reason we have explored the impact of this study on the economic analysis results in a scenario analysis (see Results of the economic analysis).

Overview

As no existing published cost-effectiveness model was considered appropriate owing to methodological limitations and non-relevance to the NHS, we developed an economic model to estimate the cost-effectiveness of second-eye cataract surgery in patients with bilateral cataracts, compared with patients with bilateral cataracts who receive only first-eye cataract surgery. The modelling was conducted following accepted standards for economic evaluation. 80,81,95

The model was structured to include outcomes proposed in the research protocol (see Appendix 1). Model data input parameters were identified from our systematic reviews of clinical effectiveness, cost-effectiveness and HRQoL, and from the wider literature where necessary. Second-eye cataract surgery is associated with a change in visual acuity and a corresponding change in HRQoL, assumed to last the patient’s lifetime. Patients undergoing surgery either experience post-surgical complications or do not. Complications are associated with a health disutility and require additional treatment and/or monitoring.

The model evaluates costs (in UK pounds using a 2012 price base) from the perspective of the NHS and Personal Social Services (PSS). Outcomes in the model are expressed as QALYs and cost-effectiveness is expressed in terms of ICERs. Uncertainty with regard to model input parameters is investigated through deterministic and probabilistic sensitivity analyses, and scenario analyses. Both costs and outcomes are discounted using a 3.5% annual discount rate in line with current guidance. 95,109

Description of the model

The conceptualisation and design of the model was informed by the assessment of the models included in our systematic review of cost-effectiveness (see Systematic review of cost-effectiveness evidence). The modelling structure was similar to that developed by Busbee and colleagues,92 with some modifications to allow patients to experience long-term complications in any subsequent year.

The UK Cataract National Dataset20 has been used to supply some of the assumptions and input parameters for the economic model, as it provides a large set of routinely collected data (55,567 cataract operations from 12 NHS trusts between 2001 and 2006), with an unselected case mix of patients considered representative of routine practice in the NHS. It reports characteristics of surgical procedures, patient demographic information and comorbidities, pre- and post-operative data on visual acuity for both first and second eyes, and intraoperative and post-operative complications. Although there are data limitations, such as the percentage of various post-operative complications being available for only 30% of all operations, the study nonetheless provides the most comprehensive published data available which is relevant to the UK and is regarded to be ‘fit for purpose’. 20

Figure 4 shows a schematic of the model. A hypothetical cohort of patients that receives second-eye cataract surgery is compared with a cohort which receives no second-eye surgery (not shown). Patients receiving surgery may experience short-term post-operative complications (endophthalmitis, cystoid macular oedema and retained lens fragments). They may also experience longer-term post-surgical complications and consequences (PCO or retinal detachment). These post-surgical complications and consequences are associated with additional outpatient visits and additional remedial procedures. Patients remain in the short- or long-term complications health states (including PCO or retinal detachment) for one cycle and then are assumed to be successfully treated. Patients may die in any model cycle (based on general population mortality rates). The model has a lifetime (25-year) horizon, with a cycle length of 1 year.

FIGURE 4.

Schematic representation of the economic model. a, post-operative complication: endophthalmitis, cystoid macular oedema, retained lens fragments; and b, long-term complications: PCO, retinal detachment.

The patient cohort starting age is 75 years based on the mean age at surgery of patients in the UK Cataract National Dataset. 20 Pre-operative visual acuity in the surgical (second) eye in the cohort was assumed to be 6/12 or better, and this was based on the UK Cataract National Dataset which reported that 55% of patients undergoing second-eye cataract surgery had a mean visual acuity of 6/12 or better in the operated eye. 20

Mean life expectancy for all patients in the model was calculated to be 9.7 years, based on UK mortality tables. 110

The model included the following assumptions:

• Utility values for the second-eye cataract surgery are assumed to remain constant over patient lifetimes, unless patients have complications from the surgery. Utility values for the no second-eye surgery group are assumed to decline over time owing to unoperated cataract progression and age-related visual acuity decline.

• Post-surgical complications are assumed to incur a disutility for 1 year.

• The model does not consider intraoperative complications as these are assumed to manifest as post-operative complications.

• It was assumed that the incidence of post-surgical complications would not differ between first- or second-eye cataract surgery and, hence, the rates used were not specific to studies of second-eye surgical patients.

• With the exception of costs for long-term post-surgical complications, the costs for outpatient visits and further procedures were assumed to be the same after the first year for the second-eye cataract surgery and no-surgery groups.

Evaluation of uncertainty

The evaluation of the cost-effectiveness of second-eye cataract surgery is subject to uncertainties about variables such as clinical effectiveness, HRQoL and resource use. This is the result of data limitations in current published studies. Uncertainty was assessed using deterministic and probabilistic sensitivity analyses (PSA). One-way deterministic sensitivity analyses were conducted to evaluate the influence of individual parameters on the model results and test the robustness of the cost-effectiveness results to variations in the structural assumptions and parameter inputs (see Results of the economic analysis). Multiparameter uncertainty in the model was addressed using PSA. 111 In the PSA, probability distributions were assigned to the point estimates used in the base case analysis. The model was run for 1000 iterations, with a different set of parameter values for each iteration, by sampling parameter values at random from their probability distributions. The uncertainty surrounding the cost-effectiveness of the treatment is represented on a cost-effectiveness acceptability curve (CEAC) according to the probability that the intervention will be cost-effective at a particular willingness-to-pay threshold.

Model validation

The model was validated by checking the model structure, calculations and data inputs for technical correctness. The structure was reviewed by clinical experts from our AG to establish that it was appropriate for the disease and its treatment. The robustness of the model to changes in input values was tested using sensitivity analyses to ensure that any changes to the input values produced changes to the results of the expected direction and magnitude. To establish its external consistency, the model results were compared with outcomes reported in clinical trials and other economic evaluations.

Data sources

Sensitivity analysis

A one-way deterministic sensitivity analysis (DSA) was performed, by varying one parameter at a time from its base case value and leaving all other variables unchanged. All parameters were varied between the high and low estimates in Table 16. The estimates used for the sensitivity analysis were based on the uncertainty of these data, and where data were available these were based on the 95% CI ranges for these parameters (e.g. resources). Where these data were not available, an alternative suitable range was chosen, based on ranges commonly used in other economic evaluations, as follows: cataract surgery costs (± 30%), post-surgical complication incidence rates (± 50%), costs for treating post-surgical complications (± 30%) and the discount rate for costs (2–5%).

Table 18 shows the results of the DSA for the model parameters. The cost-effectiveness results are robust to changes in these parameters. The model results are most sensitive to the utility gain, where the ICER varies between £1185 and £6342 per QALY. This reflects the uncertainty around the utility gain estimate from the Hiratsuka and colleagues94 study which had a 95% CI between 0.017 and 0.14. The model results are also sensitive to the cost of the cataract operation. The model results are insensitive to changes in the parameter values for the complications associated with cataract surgery.

Probabilistic sensitivity analysis

In the PSA, all parameters were sampled probabilistically from an appropriate distribution using similar ranges as used in the DSA. The parameters included in the PSA, the form of distribution used for sampling each parameter, and the upper and lower limits assumed for each variable are provided (see Appendix 10).

One thousand simulations were run. The PSA results are presented in Table 19 and are similar to the results of the deterministic analysis, with an ICER of £1970. The scatterplot for cost and QALYs for the PSA is shown in Figure 5. The CEAC is shown in Figure 6 and indicates that, at the £20,000 willingness-to-pay threshold, second-eye surgery has the highest probability of being cost-effective of 100%. The cost-effectiveness estimate was less than £10,000 per QALY in 100% of the simulations.

FIGURE 5.

Scatterplot of the additional costs and QALYs for second-eye surgery compared with no surgery.

FIGURE 6.

Cost-effectiveness acceptability curve from the PSA.

Scenario analyses with unadjusted utility gain

In addition to the sensitivity analyses, six scenario analyses were undertaken to investigate the uncertainty around structural model assumptions and input parameter values (scenarios A to F) (Table 20). A further three scenario analyses to explore the effect of adjusting utility values (scenarios G and H) and the impact of varying pre-operative visual acuity (scenario I) are described below (see Scenario analyses with adjusted utility gain).

Scenario analyses with adjusted utility gain

In the base case analysis we assumed that the utility gain estimated by Hiratsuka and colleagues94 was generalisable to the UK patient population. However, the baseline visual acuity of the patient population in Hiratsuka and colleagues’ study94 may not reflect that of UK patients. Therefore, we conducted further scenario analyses in which we adjusted the utility gain to account for this disparity in visual acuity.

We assumed that the utility gain from second-eye surgery is proportional to the improvement in visual acuity, and therefore developed a method to link visual acuity gain to utility gain. Using this method we were able to calculate the estimated utility gain according to the baseline and post-operative visual acuity.

The visual acuity of the study populations in Hiratsuka and colleagues’ study94 and in the UK Cataract National Dataset20 are shown in Table 21. Hiratsuka and colleagues94 reported their results using the log-MAR scale, whereas the UK Cataract National Dataset20 has used the Snellen scale. For the purposes of the model, we have converted the values from these studies to the decimal scale. All values shown are the visual acuity for monocular vision.

There was some uncertainty around the visual acuity gain for the second-eye cataract surgery in the Hiratsuka and colleagues study. 94 The values given in Table 21 for pre- and post-operative visual acuity are for the whole study sample, which also included patients undergoing first-eye cataract surgery and patients receiving bilateral cataract surgery. It is unclear what the post-surgical visual acuity would be if it was restricted to the subgroup of patients undergoing second-eye cataract surgery. Furthermore, we assume, for the subgroup of patients in the study undergoing second-eye surgery, that the ‘better eye’ is the eye to be operated on. This was because the study also included patients receiving first-eye surgery and we assumed that the first-eye surgery would be on the ‘worse eye’.

The mean baseline visual acuity of patients in the study by Hiratsuka and colleagues94 was log-MAR 0.16 (decimal 0.7) in the better eye. For the better eye, utility values from Hiratsuka and colleagues94 are associated with a post-surgery mean visual acuity of −0.05 log-MAR (approximate to decimal visual acuity 1.1). There was therefore an estimated visual acuity gain of 0.4 on the decimal scale.

We then calculated the expected utility gain for our modelled cohort, which represents a hypothetical cohort treated in the UK. The median baseline visual acuity of patients in the UK Cataract National Dataset20 (UKNCD) was 6/12 (decimal 0.5) and the median post-operative visual acuity for this group was 6/7.5 (decimal 0.8) and, therefore, the estimated visual acuity gain is 0.3 on the decimal scale.

The adjusted utility gain was as follows:

and the adjusted utility gain is 0.08 × 0.3/0.4 = 0.06.

Clinical effectiveness

Our systematic review demonstrates that the evidence base for the clinical effectiveness of second-eye cataract surgery is limited, with only three RCTs having met the inclusion criteria. 83–85 These RCTs appear relevant to a rather specific population group, namely elderly people in their 70s who had bilateral cataracts and had good binocular visual acuity after their first cataract operation. Although this population is relevant to the NHS, it is unclear whether it would also be representative of population groups such as those younger or older than their 70s who are eligible for second-eye cataract surgery or those with poorer visual acuity after first-eye surgery. The included RCTs are also now relatively old, having been conducted between 9 and 19 years prior to our systematic review, and it may be questioned whether or not they remain fully relevant to current clinical practice. The oldest of the RCTs was conducted by Laidlaw and colleagues84 during 1994–5, before the introduction of phacoemulsification in the UK. Compared with extracapsular cataract surgery, which was previously the standard approach, phacoemulsification has been associated with improvements in the safety of cataract surgery. 88

The RCTs83–85 all assessed effects of second-eye cataract surgery on measures of visual acuity, contrast sensitivity and stereopsis, but did not measure other vision outcomes that are affected by cataracts, such as glare disability. According to the Royal College of Ophthalmologists, glare, contrast sensitivity and patient-reported visual disability are outcomes that are increasingly being considered when making recommendations for cataract surgery. 1 A range of patient-reported outcomes covering general health and quality of life, symptoms, psychological distress and vision-related disability were reported in the included RCTs. However, several of the instruments were self-developed ad hoc by the RCT investigators without wider validation. As their clinical interpretation and reliability are unclear, we did not include these study-specific instruments in data synthesis. Where validated patient-reported measures were employed in the RCTs, there are a number of limitations and uncertainties about their interpretation (see Uncertainties).

Results of the RCTs83–85 should be interpreted with caution as they were judged to be at high risk of performance bias (although it was not practicable to mask patients to their allocated study group, so this source of potential bias was not avoidable). Assuming that the outcomes were reliable, improvements in binocular visual acuity following second-eye compared with first-eye cataract surgery would be less than 1 Snellen line equivalent, which is smaller than would normally be regarded as clinically important. However, on average, patients already had near-normal visual acuity after first-eye surgery (0.02–0.22 log-MAR, or Snellen 6/6 to 6/9.5), limiting the degree to which the second-eye surgery could have further improved their visual acuity. This ceiling effect was also evident for other outcomes reported in the RCTs, with relatively high proportions of patients also having good contrast sensitivity as well as relatively high scores for falls efficacy, handicap and vision-related disability before second-eye surgery (see Clinical effectiveness outcomes).

Among the visual outcomes assessed in the RCTs, notable improvements in stereopsis occurred following second-eye surgery, with the changes likely to be clinically meaningful to patients. Good stereopsis is important for everyday tasks that involve judging distances,35 including driving. 40 Deterioration of stereopsis can adversely affect motor skills (e.g. threading a needle, climbing stairs)42 and increases the risk of accidents such as falls,43 negatively affecting quality of life. 40 Although the evidence base is limited, stereopsis may be a more sensitive than binocular visual acuity as an indicator of patients’ vision before and after second-eye cataract surgery. However, stereopsis is unlikely to be assessed routinely in clinical practice as the stereotests required are more complex than tests of visual acuity.

The severity and type of cataract were not explicitly reported in the RCTs, but patients would appear to have had relatively mild cataracts, according to the vision parameters reported prior to second-eye surgery. A study in Singapore has suggested that the benefits of second-eye cataract surgery are related to the difference in vision between the pseudophakic and phakic eyes. 107 People who have more severe bilateral cataracts might show a greater incremental benefit from having second-eye surgery than the included RCTs suggest, as treatment of the first cataract may create a vision imbalance (e.g. anisometropia or binocular inhibition) (see Description of the underlying health problem).

Cost-effectiveness

There is a limited evidence base for the cost-effectiveness of second-eye cataract surgery. We identified only three published full economic evaluations, which varied in their study methods, patient characteristics and findings. Of these evaluations, one was a modelling study, one was a within-trial analysis and the other was an analysis alongside a prospective HRQoL study. All three had limitations, for example only one study included post-surgical complications. None was based on a systematic review of the evidence (though one was conducted alongside a RCT) or were considered wholly relevant to the NHS. For this reason it was considered necessary to develop a de novo economic model to assess cost-effectiveness.

The findings show that second-eye cataract surgery is cost-effective under conventional willingness-to-pay thresholds. The ICER in the base case analysis was £1964 per QALY gained, which is comparable with a number of other health-care interventions regarded to be cost-effective. Our results are comparable to those from the Busbee and colleagues92 study, which reported an ICER of US$2495 per QALY, but differ significantly from the other two cost-effectiveness studies. 90,91 The reason for these differences is largely driven by the utility gain for second-eye surgery assumed in the studies, and the assumptions used for long-term utility for non-second-eye cataract surgery patients. The patient cohort modelled in this study reflected the characteristics of the UK Cataract National Dataset,20 specifically the degree of visual impairment, whereas patients in the Sach and colleagues91 study were selected for inclusion in a RCT and had good vision prior to surgery.

Varying the input parameter values in the sensitivity analyses and considering alternative model scenarios did not significantly change this finding, except for one scenario varying utility values using a study that found a utility loss associated with second-eye surgery. Although there was some uncertainty around the utility change associated with second-eye surgery, with some studies in our review finding a utility gain while other a utility loss, our critical assessment of the studies found that the most appropriate HRQoL studies reported a utility gain associated with second-eye surgery.

4.1 Research aim and objectives

The aim of this project is to assess the clinical effectiveness and cost-effectiveness of second-eye cataract surgery. The objectives are:

• To conduct a systematic review of studies assessing the clinical effectiveness of second-eye surgery.

• To conduct an economic evaluation comprising: a systematic review of cost-effectiveness studies of second-eye surgery; and to develop/adapt an economic model to estimate cost-effectiveness.

Cataract removal surgery is a common procedure in the UK and has been shown to be cost-effective in the initial eye. 1 Some patients with bilateral cataract may only have surgery on one eye, but it is suggested that surgery on the second eye may have additional benefit for patients in terms of improving vision and being able to perform everyday activities (e.g. being able to drive). However, there is debate about how cost-effective second-eye surgery would be.

Scoping searches for this protocol have identified three published randomised controlled trials (RCTs) of second-eye cataract surgery compared with first-eye surgery. 2–4 All of these trials reported varying degrees of benefit associated with second-eye surgery, in terms of improved visual acuity, visual symptoms, and quality of life. No published systematic reviews of the clinical effectiveness and cost-effectiveness of second-eye cataract surgery have been identified. Three published economic evaluations have been identified5–7 (one of which was a trial-based evaluation conducted in the UK5) which have used divergent methods and generated mixed results (see section 6). An evidence synthesis and economic evaluation would therefore be useful to inform health service policy and practice in this area.

4.2 Background

A cataract is a clouding that develops in the crystalline lens of the eye which results in symptoms such as blurred or reduced vision, and problems associated with glare or low-contrast conditions. Formation of cataracts is linked to ageing, with the gradual accumulation of yellow-brown pigment within the lens reducing light transmission. Cataracts can also be congenital or secondary to other causes (e.g. chronic uveitis). Other, extrinsic, risk factors include smoking, alcohol diabetes, and use of systemic corticosteroids. In developing countries malnutrition and acute dehydrating diseases are also associated with cataract development. 8

Cataracts are a common cause of visual impairment worldwide, and are more common in older people. The North London Eye Study randomly sampled people aged over 65 years from general practices in north London (1547/1840 responding), and found a prevalence of cataract causing visual impairment (visual acuity in one or both eyes poorer than 6/12, attributable to lens opacity) of 30%. 9 The prevalence of cataract increased steadily with age from 16% in the 65 to 69 age group, to 71% in people aged over 85 years. In the MRC Trial of Assessment and Management of Older People in the Community, nurses tested visual acuity in 14,403 people aged 75 years and older from 49 general practices in Britain. 10 Of 976 people with binocular visual impairment (binocular acuity <6/18) excluding refractive error, 36% were classified as having cataract.

4.3 Definition of the intervention

Cataract extraction surgery is considered to be the only curative intervention available. First-eye cataract surgery refers to removal of the cataract in only one eye. Second-eye cataract surgery is performed in patients with bilateral cataract at a point in time following first-eye surgery. Surgery involves removal of the natural (crystalline) lens of the eye that has developed an opacification (cataract) and replacement with a synthetic lens to restore clarity of vision.

Phacoemulsification is the standard method of cataract removal in the NHS and is associated with better visual outcome than conventional extracapsular surgery. 11 It involves making small incisions (e.g. 3 mm in width) where the clear front covering (cornea) meets the white of the eye (sclera). A circular opening is created on the lens surface (capsule). A small surgical instrument (phaco probe) is inserted into the eye and ultrasound waves are used to break the cataract into small pieces. The cataract and lens pieces are removed from the eye using suction and an intraocular lens implant may then be placed inside the lens capsule. The procedure is usually carried out under local anaesthetic as day surgery.

4.4 Place of the intervention in the treatment pathway(s)

People with cataracts may be referred to an ophthalmologist by a GP or optometrist. Referral criteria include clinically significant visual symptoms related to the cataract (e.g. reduced visual acuity, functional impairment), negative effects on the patient’s lifestyle by the cataract, and patient’s wish to undergo surgery.

Cataract extraction is one of the most common elective surgical procedures in the UK. In 2009–10 there were 334,142 cataract extractions performed in England. 12 It is estimated that over a one-third of NHS cataract operations are performed in the second eye. 13 Patients are generally assessed for their second-eye surgery at their first-eye surgery post-operation check (e.g. 4 weeks post-operation). They then may wait up to a further 18 weeks for surgery unless there is an urgent reason to do it sooner. It is very uncommon for cataract extraction to take place in both eyes simultaneously due to the risk of associated complications including bilateral infective endophthalmitis (inflammation of the inside of the eye) which may lead to blindness in both eyes. Expert clinical opinion suggests that 1% or fewer extractions are simultaneous.

The indication for second-eye surgery is based on whether the patient has a cataract in the second eye, is symptomatic and wishes to undergo surgery. Patients would have a glasses check so that they can assess the full benefit of the first-eye surgery, in order to correctly judge their level of symptoms and visual rehabilitation with a change in glasses. The threshold for impairment to visual acuity would be the same as for initial cataract extraction surgery, and expert clinical opinion suggests that thresholds may vary between health trusts. A proportion of patients choose not to undergo second-eye surgery and some die before they develop a second cataract.

The Royal College of Ophthalmologists cataract surgery guidelines (2010) state that it is clinically and economically appropriate for second-eye surgery to be offered to those patients who want it. A 1b recommendation is given (recommendation based on at least one randomised trial). 13 Similarly, the Scottish Health Technologies Group issued an advice statement in September 2012 advising that there is RCT evidence to support second-eye surgery and evidence from cost–utility analysis to demonstrate lifetime cost-effectiveness. 14 In England and Wales, the National Institute for Health and Care Excellence (NICE) has not appraised second-eye cataract surgery.

4.5 Relevant comparators

Single eye cataract surgery can be considered a relevant comparator in patients with bilateral cataracts not currently scheduled for second-eye surgery. They may receive additional supportive care such as prescription glasses.

4.5 Outcomes

The clinical effectiveness of cataract surgery can be measured in a variety of ways. Commonly used clinical visual measures include: visual acuity (clearness of vision measured via, for example, a Snellen chart); stereopsis (depth perception); contrast sensitivity (the ability to see objects that may not be outlined clearly or that do not stand out from their background) and stereoacuity (the smallest detectable depth difference that can be seen in binocular vision).

Functional status is measured using assessment tools including the visual functioning index (VF-14), a well-established patient questionnaire designed to measure functional impairment caused by cataracts. In the VF-14 patients rate their ability to undertake activities such as reading a newspaper or a book, driving and reading traffic signs or taking part in games, and a total score is computed representing the degree of visual function. 15 It is suggested, however, that the VF-14 is not sensitive to all cataract symptoms and therefore a new instrument to assess the impact of cataract surgery in terms of visual function and quality of life is needed. 16

The health-related quality of life of people with cataracts has been assessed in clinical trials using generic instruments such as the SF-362 and the EuroQoL EQ-5D. 5 The EQ-5D has five dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) with three possible levels of severity for each.

Like all forms of surgery, cataract removal is associated with adverse events. These can include: endophthalmitis; retinal detachment; bullous keratopathy (swelling of the cornea); and intraocular lens dislocation.

4.6 Population and relevant subgroups

As discussed earlier, cataracts mainly affect older people, with increasing prevalence with age. Some people may have eye comorbidities such as age-related macular degeneration or glaucoma.

5.1 Search strategy

A comprehensive search strategy will be developed, tested and refined by an experienced information scientist (see Appendix for draft MEDLINE search strategy). Separate searches will be conducted to identify studies of clinical effectiveness, cost-effectiveness, Health Related Quality of Life (HRQoL), resource use and costs, and epidemiology.

The search strategy will comprise the following main elements:

• Searching of electronic databases

• Contact with experts in the field

• Scrutiny of bibliographies of retrieved papers

Electronic databases to be searched will include:

• General health and biomedical databases – MEDLINE (Ovid); PreMedline In-Process & Other Non-Indexed Citations; EMBASE; the Cochrane Central Register of Controlled Trials; and the Science Citation Index.

• Specialist databases – the Cochrane Database of Systematic Reviews; Database of Abstracts of Reviews of Effectiveness (DARE); Health Technology Assessment database; EconLit; NHS Economic Evaluation Database.

• Grey literature and research in progress – UK Clinical Research Network Portfolio Database; and Conference Proceedings Citation Index –Science (Web of Science); Current Controlled Trials; Clinical Trials.gov; BIOSIS; NIHR Clinical Research Network Portfolio; CenterWatch; Health Services Research Projects in Progress and Computer Retrieval of Information on Scientific Projects (CRISP).

All databases will be searched from inception to the current date and searches will be limited to English language.

5.2 Inclusion/exclusion criteria

• Population: adults (aged 18 or over) who have had one cataract operation already and still have or develop significant cataract causing visual impairment in the other eye.

• Interventions: cataract surgery for the second eye. Studies reporting any surgical technique will be included.

• Comparators: cataract extraction surgery in one eye only.

• Outcomes: clinical visual measures (visual acuity; stereoacuity; contrast sensitivity); patient-reported visual disability and symptoms (e.g. VF-14); patient satisfaction with surgery and vision; health-related quality of life (e.g. EQ-5D); adverse events.

• Types of studies: Randomised controlled trials (RCTs) will be included. If necessary non-RCT data will be sought to inform the cost-effectiveness analysis (e.g. on safety). Any systematic reviews identified will be used only as a source of references.

• Studies published as abstracts or conference presentations will only be included if sufficient details are presented to allow an appraisal of the methodology and the assessment of results to be undertaken.

5.3 Inclusion, data extraction and quality assessment process

Studies will be selected for inclusion through a two-stage process using the predefined and explicit criteria (as specified in section 5.2). The literature search results will be screened by two reviewers to identify all citations that may meet the inclusion criteria. Full manuscripts of relevant studies will be retrieved and assessed by two reviewers using a standardised eligibility form.

Data extraction and quality assessment will be undertaken by one reviewer and checked by a second reviewer using a pre-designed and piloted data extraction form to avoid any errors. At each stage, any disagreements between reviewers will be resolved by consensus or if necessary by arbitration by a third reviewer.

5.4 Quality assessment

Included trials will be assessed in terms of their risk of bias (e.g. selection bias, detection bias, performance bias, attrition bias, and selective reporting bias) using Cochrane Collaboration criteria. 19,20 Aspects of study quality including statistical procedures, outcome measurement and generalisability will also be assessed.

5.5 Methods of analysis/synthesis

Studies will be synthesised through a structured narrative review with tabulation of results of included studies. Where appropriate and where suitable data are available, meta-analysis will be employed to estimate a summary measure of effect on relevant outcomes. The specific methods for meta-analysis and for the detection and investigation of heterogeneity will depend upon the summary measure selected and will used standard procedures recommended by the Cochrane Collaboration. 19 Cochrane Review Manager (RevMan) software will be used to perform any meta-analysis. Heterogeneity will be explored through consideration of the study populations, methods and interventions, by visualisation of results and, in statistical terms, by the χ² test for homogeneity and the I² statistic.

6.1 Review of published cost-effectiveness studies

The sources detailed in section 5.1 will be used to identify studies of the cost-effectiveness of second-eye cataract surgery. Studies will be included in the systematic review of cost-effectiveness if they are full economic evaluations (cost-effectiveness, cost–utility or cost–benefit analyses) that report both measures of costs and consequences. The methodological quality of included studies will be assessed using accepted criteria for appraising economic evaluations. 21 Where relevant this will be supplemented with additional criteria for critical appraisal of model-based evaluations. 22 Studies will be synthesised through a narrative review that includes a clear explanation of the assessment process, detailed critical appraisal of study methods, critical assessment of data used in any economic models and tabulation of the results of included studies.

Published studies conducted in the UK and adopting an NHS and Personal Social Services (PSS) perspective will be examined in more detail. Stand alone cost analyses based in the UK NHS will also be searched for – these will not be included in the systematic review, but will be retained as sources of information on resource use and cost associated with second-eye cataract surgery (including short-term and longer-term adverse events).

Scoping searches have identified two published cost–utility studies of second-eye cataract surgery,5,7 in addition to the one study identified in the NIHR HTA programme commissioning brief, by Busbee and colleagues. 6 One of the additional studies was conducted in the UK and has an NHS and Personal Social Services (PSS) perspective. 5 Neither of the additional studies is a model-based evaluation – one was a study of ‘routine’ cataract surgery (of 219 patients, 73 had both eyes operated on and 59 had a second-eye operation, the first eye having been operated earlier)7, the other was conducted alongside a RCT. 5

6.2 Evaluation of costs and cost-effectiveness

Existing economic models developed to estimate the cost-effectiveness of second-eye cataract surgery, identified in the systematic review of economic evaluations, will be assessed for their quality, relevance and suitability for adoption in the current review. If considered relevant and valid the models will be adapted (if required) and populated with updated (and UK-practice-relevant) clinical and cost parameter values using data identified in our clinical and cost-effectiveness reviews.

The decision tree model presented by Busbee and colleagues,6 will be considered for adaptation in this economic evaluation. Key assumptions of this study will be discussed with our clinical and methodological advisors for their appropriateness. These include:

• the inclusion and timing of adverse events: endophthalmitis; cystoid macular oedema; lost lens fragments (each occurring within 4 months of surgery); posterior capsular opacification (occurring at rate of 28% over 5 years, with treatment occurring on average 2 years after surgery); retinal detachment (in 0.81% of cases, with treatment on average 1 year after surgery); intraocular lens dislocation (in 1.1% of cases); and pseudophakic bullous keratopathy (in 0.3% of cases, with treatment on average 1 year after surgery). Retinal detachment following posterior capsular opacification was assumed to occur in 3% of cases.

• the most appropriate measure of outcome. The model uses assumed levels of bilateral visual acuity which are then mapped to utility values derived using the time-trade-off technique. However, it is recognised that bilateral visual acuity may have limitations and other measures, including patient-reported visual dysfunction, may be more informative.

Current guidelines for good practice in decision-analytic modelling and the general principles outlined in the NICE ‘reference case’ will be followed. 22–24 Development of the structure and parameters of the model will be informed by several sources including previous models identified in the systematic review of cost-effectiveness, evidence from our systematic review of clinical effectiveness, as well as guidance from clinical and methodological advisors. The model will be validated through discussion with expert advisors. Additional targeted literature searches will be required to populate other parameters in the model as necessary.

The model will adopt a UK NHS and PSS perspective with cost and outcomes discounted at an annual rate of 3.5%. The model will present estimates of the cost-effectiveness of second-eye cataract surgery, in terms of incremental cost per quality adjusted life-year (QALY) gained, compared with first-eye surgery only.

Resource use for second-eye cataract surgery, including management of adverse events, will be estimated from studies included in the systematic review of clinical effectiveness, published costing studies identified by our searches, any relevant clinical guidelines and from discussion with expert advisors.

As far as is possible, costings developed for the model will proceed by first identifying and quantifying resource use and then applying appropriate unit costs. Where resource use data from published literature is insufficient we would use estimates from relevant clinical experts and this will be clearly identified in the final report. To develop unit cost estimates we will assess official, nationally representative sources (NHS Reference Costs,25 Unit Costs of Health and Social Care,26 British National Formulary27) for applicability and level of detail, as well as unit cost estimates applied in studies included in the systematic review of cost-effectiveness and in costing studies identified by our searches. If these sources are inadequate we would develop unit cost estimates in collaboration with the costing unit at Southampton University Hospitals NHS Trust. Costs will be inflated to current prices using the Hospital and Community Health Services Pay and Prices Index, where necessary. 26

Health-related quality-of-life (HRQoL) data, where available, will be extracted from studies included in the clinical- and cost-effectiveness systematic reviews. Where available, the impact of treatment adverse effects on patients will also be incorporated. Where QoL data are insufficient to calculate utility estimates, data will be derived from the broader literature or estimated from other sources. In accordance with the NICE methodological guide for technology appraisals,24 the utility values used in the model will be elicited where possible from the general population using a preference-based method. Where these are not available, utility estimates will be derived from alternative sources and the assumptions made will be explicitly stated.

The cost–utility studies identified in our preliminary searches vary in the approaches used to estimate the utility gain associated with second-eye cataract surgery. Busbee and colleagues6 estimated levels of bilateral visual acuity (they assumed the same utility values based on VA outcomes for patients having first-eye surgery) and mapped these to previously estimated utility values associated with given levels of bilateral visual acuity. This yielded comparatively large utility gains of 0.109. In contrast, the other studies based their utility estimates on patient responses to generic HRQoL instruments (EQ-5D5 and 15D7) valued using population-derived tariffs. These both estimated substantially lower utility gains following second-eye surgery [a small QALY gain of 0.015 for a 12 month time horizon (difference in EQ-5D utility at post-surgery follow-up not reported) in one study6 and a statistically non-significant decrease of −0.01 in the other7]. Both studies using patient-reported QoL measures included patients with limited vision loss (Sach and colleagues5 report that 86% of trial participants had baseline visual acuity of 6/12 or better, while Rasanen and colleagues7 noted that 17% of patients reported having no pre-operative difficulty in seeing and 47% only minor difficulties). It is not clear how far differences in patient populations studied may have given rise to divergent results in these studies, although baseline visual dysfunction (in the eye to be operated on) is likely to be an important factor in determining the potential gain from second-eye surgery. Current Cataract Surgery Guidelines published by the Royal College of Ophthalmologists13 note potential benefits from second-eye surgery, but do not indicate the degree of visual dysfunction in the second eye at which it would be appropriate to undertake second-eye surgery.

Sensitivity analyses and scenario analyses will be conducted with respect to variables over which there is greatest uncertainty. For the deterministic analyses this will be oriented towards variables with the greatest uncertainty over their methods of derivation or where choices/judgments have had to be made between alternative sources. The key variables to be explored in sensitivity/scenario analyses are likely to be the clinical benefit (for example gain in bilateral visual acuity) from second-eye cataract surgery and the utility associated with such benefits. However the robustness of model results to other clinical variables (including incidence and timing of adverse events) and to resource use assumptions will also be considered. The importance of the underlying model assumptions will be assessed through an analysis of different scenarios, particularly where evidence to populate the model is inadequate or conflicting (for example where the model uses data derived using expert opinion).

The results of the probabilistic sensitivity analysis will be presented using cost-effectiveness acceptability curves (CEACs).

Advisory group

An AG has been recruited comprising clinical experts in ophthalmology, experts in health technology assessment methodology (including health economics) and representatives from patient organisations. The group will has commented on the draft protocol and will comment on the draft final report. The group will be consulted during the course of the project for advice as necessary. The current members of the group are:

• Professor Janet Marsden, Professor of Ophthalmology and Emergency Care & Chair of the RCN Ophthalmic Nursing Forum, Research Institute for Health and Social Change (Health Care Studies Department), Manchester Metropolitan University.

• Professor John Sparrow, Consultant Ophthalmologist, University Hospitals Bristol NHS Foundation Trust, & Honorary Professor of Ophthalmic Health Services Research and Applied Epidemiology, University of Bristol.

• Mr Simon P Kelly, Chair of Quality & Safety subcommittee of the Royal College of Ophthalmologists, & Consultant Ophthalmic Surgeon at Royal Bolton Hospitals NHS Foundation Trust.

• Dr Ewen Cummins, Health Economist, McMaster Development Consultants

Additional members may be recruited to the group during the course of the project, as required.

10.1. Draft Medline (Ovid) search strategy

1. exp cataract extraction/ (26,455)

2. (cataract* adj5 (surg* or remov* or extract* or procedure* or operat* or excis* or aspirat* or incis*)).tw. (19,067)

3. phacoemulsification.tw. (5253)

4. (PKE or PCIOL or ECCE or ICCE or MSICS or MISICS or SICS).tw. (931)

5. (pseudoaphakia or pseudoaphakic or phakectomy or phakectomies or “enzymatic zonulolysis” or “zonulolyses enzymatic” or “enzymatic zonulolyses” or “zonulolysis enzymatic”).tw. (107)

6. lens implantation, intraocular/ or lenses intraocular/ (16,135)

7. “cataract patient*”.tw. (1092)

8. or/1-7 (36,944)

9. (“single eye*” or “right eye*” or “left eye*” or “one eye” or “dominant eye*” or “better seeing eye*” or “either eye” or “unilateral cataract*” or “first eye*”).tw. (22,936)

10. (“fellow eye*” or “second eye*” or “both eyes” or “other eye*” or “two eyes” or “next eye*”).tw. (16,331)

11. (eye* adj5 (sequential* or simultaneous* or serial*)).tw. (1275)

12. (bilateral* adj5 cataract*).tw. (1369)

13. 8 and (9 or 10 or 11 or 12) (3749)

14. Randomized Controlled Trials as Topic/ (83,981)

15. randomized controlled trial.pt. (339,605)

16. controlled clinical trial.pt. (85,425)

17. Controlled Clinical Trial/ (85,425)

18. placebos/ (31,477)

19. random allocation/ (76,252)

20. Double-Blind Method/ (117,819)

21. Single-Blind Method/ (16,898)

22. (random* adj2 allocat*).tw. (18,020)

23. placebo*.tw. (140,190)

24. ((singl* or doubl* or trebl* or tripl*) adj (blind* or mask*)).tw. (115,464)

25. crossover studies/ (30,752)

26. (crossover* or (cross adj over*)).tw. (51,975)

27. Research Design/ (68,167)

28. ((random* or control*) adj5 (trial* or stud*)).tw. (452,653)

29. Clinical Trials as Topic/ (163,152)

30. random*.ab. (567,167)

31. or/14-30 (1,175,731)

32. 13 and 31 (612)

1.1 Health technology

1.2 Interventions and comparators

What interventions/strategies were included?

Was a no treatment/supportive care strategy included?

Describe interventions/strategies

1.3 Research question

What are the stated objectives of the evaluation?

1.4 Study type

Cost-effectiveness/cost–utility/cost–benefit analysis?

1.5 Study population

What definition was used for [condition]? What are the characteristics of the baseline cohort for the evaluation?

1.6 Institutional setting

Where is/are the intervention(s) being evaluated usually provided?

1.7 Country/currency

Has a country setting been provided for the evaluation? What currency are costs expressed in and does the publication give the base year to which those costs relate?

1.8 Funding source

1.9 Analytical perspective

What is the perspective adopted for the evaluation (health service, health and Personal Social Services, third-party payer, societal (i.e. including costs borne by individuals and lost productivity)?

3.1 Indirect costs (costs due to lost productivity, unpaid inputs to patient care)

Were indirect costs included?

4.1 List the utility values used in the evaluation

5.1 Extract transition probabilities for [natural history/disease progression] model and show sources (or refer to table in text)

5.2 What is the model time horizon?

5.3 What, if any, discount rates have been applied in the model? Same rate for costs and outcomes?

6.1 Provide a summary of the clinical outcome/benefits estimated for each intervention/strategy assessed in the evaluation

6.2 Provide a summary of the costs estimated for each intervention/strategy assessed in the evaluation

6.3 Synthesis of costs and benefits – are the costs and outcomes reported together (e.g. as cost-effectiveness ratios)? If so, provide a summary of the results

6.4 Give results of any statistical analysis of the results of the evaluation

6.5 Was any sensitivity analysis performed – if yes, what type(s) [i.e. deterministic (one-way, two-way, etc.) or probabilistic]

6.6 What scenarios were tested in the sensitivity analysis? How do these relate to structural uncertainty (testing assumptions over model structure such as relationships between health states), methodological uncertainty (such as choices of discount rate or inclusion of indirect costs) or parameter uncertainty (assumptions over values of parameters in the model, such as costs, quality of life or disease progression rates)?

6.7 Give a summary of the results of the sensitivity analysis – did they differ substantially from the base case analysis. If so, what were the suggested causes?

7.1 What are the implications of the evaluation for practice?

1.1 Health technology

1.2 Interventions and comparators

What interventions/strategies were included?

Was a no treatment/supportive care strategy included?

Describe interventions/strategies.

1.3 Research question

What are the stated objectives of the evaluation?

1.4 Study type cost-effectiveness/cost–utility/cost–benefit analysis?

1.5 Study population

What definition was used for [condition]? What are the characteristics of the baseline cohort for the evaluation?

1.6 Institutional setting where is/are the intervention(s) being evaluated usually provided?

1.7 Country/currency

Has a country setting been provided for the evaluation? What currency are costs expressed in and does the publication give the base year to which those costs relate?

1.8 Funding source

1.9 Analytical perspective

What is the perspective adopted for the evaluation (health service, health and personal social services, third-party payer, societal (i.e. including costs borne by individuals and lost productivity)?

3.1 Indirect costs (costs due to lost productivity, unpaid inputs to patient care)

Were indirect costs included:

4.1 List the utility values used in the evaluation

Indicate the source for individual cost values (if appropriate).

5.1 Extract transition probabilities for [natural history/disease progression] model and show sources (or refer to table in text)

5.2 What is the model time horizon?

5.3 What, if any, discount rates have been applied in the model? Same rate for costs and outcomes?

6.1 Provide a summary of the clinical outcome/benefits estimated for each intervention/strategy assessed in the evaluation

6.2 Provide a summary of the costs estimated for each intervention/strategy assessed in the evaluation

6.3 Synthesis of costs and benefits – are the costs and outcomes reported together (e.g. as cost-effectiveness ratios)? If so, provide a summary of the results

6.4 Give results of any statistical analysis of the results of the evaluation

6.5 Was any sensitivity analysis performed – if yes, what type(s) (i.e. deterministic (one-way, two-way, etc.) or probabilistic)

6.6 What scenarios were tested in the sensitivity analysis? How do these relate to structural uncertainty (testing assumptions over model structure such as relationships between health states), methodological uncertainty (such as choices of discount rate or inclusion of indirect costs) or parameter uncertainty (assumptions over values of parameters in the model, such as costs, quality of life or disease progression rates)?

6.7 Give a summary of the results of the sensitivity analysis – did they differ substantially from the base case analysis. If so, what were the suggested causes?

7.1 What are the implications of the evaluation for practice?

1.1 Health technology

1.2 Interventions and comparators

What interventions/strategies were included?

Was a no treatment/supportive care strategy included?

Describe interventions/strategies.

1.3 Research question

What are the stated objectives of the evaluation?

1.4 Study type cost-effectiveness/cost–utility/cost–benefit analysis?

1.5 Study population

What definition was used for [condition]? What are the characteristics of the baseline cohort for the evaluation?

1.6 Institutional setting where is/are the intervention(s) being evaluated usually provided?

1.7 Country/currency

Has a country setting been provided for the evaluation? What currency are costs expressed in and does the publication give the base year to which those costs relate?

1.8 Funding source

1.9 Analytical perspective

What is the perspective adopted for the evaluation (health service, health and PSS, third-party payer, societal (i.e. including costs borne by individuals and lost productivity)?

3.1 Indirect costs (costs due to lost productivity, unpaid inputs to patient care)

Were indirect costs included:

4.1 List the utility values used in the evaluation

5.1 Extract transition probabilities for [natural history/disease progression] model and show sources (or refer to table in text)

5.2 What is the model time horizon?

5.3 What, if any, discount rates have been applied in the model? Same rate for costs and outcomes?

6.1 Provide a summary of the clinical outcome/benefits estimated for each intervention/strategy assessed in the evaluation

6.2 Provide a summary of the costs estimated for each intervention/strategy assessed in the evaluation

6.3 Synthesis of costs and benefits – are the costs and outcomes reported together (e.g. as cost-effectiveness ratios)? If so, provide a summary of the results

6.4 Give results of any statistical analysis of the results of the evaluation

6.5 Was any sensitivity analysis performed – if yes, what type(s) (i.e. deterministic (one-way, two-way, etc.) or probabilistic)

6.6 What scenarios were tested in the sensitivity analysis? How do these relate to structural uncertainty (testing assumptions over model structure such as relationships between health states), methodological uncertainty (such as choices of discount rate or inclusion of indirect costs) or parameter uncertainty (assumptions over values of parameters in the model, such as costs, quality of life or disease progression rates)?

6.7 Give a summary of the results of the sensitivity analysis – did they differ substantially from the base case analysis. If so, what were the suggested causes?

7.1 What are the implications of the evaluation for practice?

Study characteristics

Data sources