Decision problem
Some patients with bilateral cataracts may have surgery on only 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 uncertainty about how cost-effective second-eye surgery would be, with no systematic reviews hitherto having addressed this question. The decision problem is relevant to current financial constraints faced by the NHS, with concerns having been raised about possible overprovision of cataract surgery55,56and the need to ensure adequate patient access to cataract surgery.58,77
Overall aims and objectives of the assessment
The aim of this systematic review was to assess the clinical effectiveness and cost-effectiveness of second-eye cataract surgery. The objectives were:
l to conduct a systematic review of studies assessing the clinical effectiveness of second-eye surgery
l to conduct an economic evaluation comprising:
¢ a systematic review of cost-effectiveness studies of second-eye surgery
¢ a systematic review of relevant quality-of-life studies
¢ an economic model, developed de novo or adapted from an existing one, to estimate cost-effectiveness of second-eye cataract surgery.
Chapter 3 Methods
T
he a priori methods for systematically reviewing the evidence of clinical effectiveness and cost-effectiveness are described in the research protocol (see Appendix 1). The protocol was sent to our expert advisory group (AG) (see Acknowledgements) for comment and minor amendments were made as appropriate. None of the comments received identified specific problems with the methods of the review.Identification of studies
Sensitive search strategies were developed and refined by an experienced information specialist (see Appendix 2).
Main searches for clinical effectiveness, cost-effectiveness and HRQoL literature were undertaken from inception of databases to March/April 2013, with updated searches conducted in July 2013. Searches were limited to the English language.
The strategies were applied to the following databases:
l Ovid MEDLINE (1946 to March 2013)
l Ovid MEDLINE Daily Update (12 March 2013)
l Ovid MEDLINE In-Process & Other Non-Indexed Citations (12 March 2013)
l Ovid EMBASE (1974–March 2013)
l Web of Science: Science Citation Index-Expanded (SCI-Expanded; 1970–March 2013); Conference Proceedings Citation Index– Science (CPCI-S; 1990–March 2013); Conference Proceedings Citation Index– Social Science and Humanities (CPCI-SSH)
l BIOSIS Previews (Web of Science platform; 1956–March 2013)
l The Cochrane Library: Cochrane Central Register of Controlled Trials; Cochrane Database of Systematic Reviews
l Centre for Reviews and Dissemination (CRD): Database of Abstracts and Reviews of Effectiveness;
Health Technology Assessment (HTA) database.
In addition to the databases listed above, the searches of cost-effectiveness and HRQoL also included the following databases:
l NHS Economic Evaluations Database (NHS EED) (cost-effectiveness searches)
l The King’s Fund database (HRQoL searches).
For each of the systematic reviews of clinical effectiveness, cost-effectiveness and HRQoL, internet pages of the Cochrane Eyes and Vision Group, the Royal College of Ophthalmologists and NICE were searched.
The reference lists of included primary studies were checked for additional references. If any relevant systematic reviews were identified, their reference lists were also checked for relevant references.
Experts on the review AG were also asked to identify additional published and unpublished references.
All search results were imported into a Reference Manager (Thomson ResearchSoft, San Francisco, CA, USA) database.
Study selection
Titles and abstracts of records identified by the bibliographic searches for clinical effectiveness literature were assessed independently by two reviewers for potential eligibility, using a standardised eligibility selection worksheet (see Appendix 3) containing the following pre-specified eligibility criteria:
l Population: adults (aged 18 years and over) who have had one cataract operation already and still have or have developed a significant cataract causing visual impairment in the other eye.
l Intervention(s): cataract surgery for the second eye. Studies reporting any surgical technique were included.
l Comparator(s): cataract surgery in one eye only (can include a‘waiting list control’ group in which the comparator population ultimately receives second-eye surgery, but not during the study period); and patients may receive additional supportive care if this is usual practice, such as prescription glasses.
l Outcomes: any clinical visual measures (including measures of visual acuity; contrast sensitivity;
stereopsis); patient-reported visual disability and symptoms (e.g. VF-14); patient satisfaction with surgery and vision; HRQoL (e.g. EQ-5D); and adverse events (including peri- and post-operative complications).
l Study design: RCTs. If relevant systematic reviews were identified, these were only used as a source of references. Studies published as abstracts or conference presentations were only included if sufficient details were presented to allow an appraisal of the methodology and the assessment of results to be undertaken.
The systematic reviews of cost-effectiveness and HRQoL employed the inclusion criteria listed above, with the following modifications (which were reflected in the study selection worksheets):
l Studies of any design were eligible for the cost-effectiveness review if they reported full economic evaluations (e.g. cost-effectiveness, cost–utility).
l Studies of any design were eligible for the systematic review of HRQoL if they used generic preference-based HRQoL measures or generic preference valuation methods, and reported HRQoL outcomes. It was anticipated that there would be limited relevant literature on HRQoL in patients receiving second-eye cataract surgery. Therefore, the criteria were broader and could also include first-eye cataract surgery. While HRQoL data based on second-eye surgery were to be
prioritised for use in the economic evaluation, data from studies of first-eye surgery could be included for context and to potentially inform sensitivity analysis.
Any disagreements between the reviewers regarding title and abstract screening using the above criteria were resolved by discussion or, if necessary, consultation with a third reviewer. Full-text records were then obtained for all titles and abstracts that met the inclusion criteria or remained unclear. The full-text records were assessed independently by two reviewers using the same criteria as for titles and abstracts.
Any further disagreements between the reviewers were resolved through discussion and, if necessary, recourse to a third independent reviewer.
For the systematic review of HRQoL, the inclusion criteria were slightly revised for screening full papers, with the additional stipulation that to be included studies had to be primary research, and had to report utility values (i.e. patient preferences for a health outcome, commonly measured on an interval scale with 0 reflecting states of health equivalent to death and 1 reflecting perfect health) as opposed to, for example, reporting only HRQoL scores (i.e. not related to patient preferences).
Only articles published in English were eligible for the systematic reviews. Conference abstracts were not eligible for the systematic review of cost-effectiveness and HRQoL but were considered for the review of clinical effectiveness, subject to reporting adequate detail to enable an appraisal of the methodology.
Data extraction and critical appraisal
Standardised forms were used to collect data from the studies that were included in the systematic reviews (completed data extraction forms for the reviews of clinical effectiveness, cost-effectiveness and HRQoL are shown, see Appendices 4, 7 and 9, respectively). In each systematic review, data were extracted by one reviewer and then checked for accuracy by a second reviewer.
Randomised controlled trials included in the systematic review of clinical effectiveness were assessed in terms of their risk of bias (selection bias, detection bias, performance bias, attrition bias and selective reporting bias) using Cochrane Collaboration criteria.78,79Risk of bias was assessed independently by two reviewers who then met to compare their judgements. Disagreements were resolved through discussion and, if necessary, consultation with a third reviewer. Tables summarising and explaining the risk of bias judgements are included in the data extraction forms (see Appendix 4). Other aspects of study quality relating to statistical procedures, outcome measurement and generalisability were also assessed and recorded in the data extraction forms.
The methodological quality of studies included in the systematic review of cost-effectiveness was assessed using accepted criteria for appraising economic evaluations.80,81Owing to the diverse types of study design eligible, the methodological quality of studies included in the systematic review of HRQoL was
not formally assessed.
Method of data synthesis
As specified in the protocol (see Appendix 1), studies were synthesised narratively following a structured approach similar to one proposed by Rodgers and colleagues.82Quantitative pooling of outcomes across clinical effectiveness studies in a meta-analysis was not considered appropriate as the included studies differed in their methodological characteristics (see Chapter 4, Clinical effectiveness outcomes).
Meta-analysis was also not possible for pooling outcomes in the systematic reviews of cost-effectiveness or HRQoL because of the different types of study design included (see Chapter 5, Economic analysis).