The measurement of visual impairment in national surveys:
a review of available data sources
Prepared for the Thomas Pocklington Trust
The measurement of visual impairment
in national surveys:
a review of available data sources
Jo d’Ardenne, Matt Hall, Sally McManus
October 2011
Prepared for the Thomas Pocklington Trust
ContentsAcknowledgements......
Executive summary ......
The Pocklington Trust......
Background......
Aims......
Methodology......
Notes only:
Results......
Discussion......
Recommendations......
1Background and approach......
1.1Background......
1.1.1Estimates of the prevalence of visual impairment......
1.1.2Factors influencing estimates of visual impairment......
1.1.3Use of general population surveys in estimating prevalence......
1.1.4Variation in the definition of visual impairment......
1.2Aims of this survey review......
1.3Process of review......
1.4Summary of methodological approach......
1.5Consultation with experts......
1.6Report structure......
2Types of data collection......
2.1National surveys of the general population......
2.2National surveys using non-random sampling......
2.3Regional surveys......
2.4Surveys of people with visual impairment......
2.5Surveys of disabled people......
2.6Surveys of other population subgroups......
2.7Surveys with objective measures......
2.8Administrative data sources......
2.9Qualitative research with visually impaired people......
3Methods......
3.1Identification of surveys to include in the analysis......
3.1.1Survey inclusion criteria......
3.1.2Eligible surveys not included in the analysis......
3.1.3Surveys included in the analysis......
3.2Analysis......
3.2.1Statistical approach......
3.2.2Weighting......
3.2.3Missing data......
4Surveys......
4.1Health Survey for England (HSE)......
4.1.1Visual impairment questions asked......
4.1.2Long-standing illness code frame......
4.1.3Visual impairment prevalence estimates......
4.1.4Methodological issues......
4.1.5Study contacts and further information......
4.2Scottish Health Survey (SHeS)......
4.2.1Visual impairment questions asked......
4.2.2Visual impairment prevalence estimates......
4.2.3Methodological issues......
4.2.4Study contacts and further information......
4.3Welsh Health Survey (WHS) 2007......
4.3.1Visual impairment questions asked......
4.3.2Visual impairment prevalence estimates......
4.3.3Methodological issues......
4.3.4Study contacts and further information......
4.4Adult Psychiatric Morbidity Survey (APMS) 2007......
4.4.1Visual impairment questions asked......
4.4.2Visual impairment prevalence estimates......
4.4.3Methodological issues......
4.4.4Study contact and further information......
4.5English Longitudinal Study of Aging (ELSA) Wave 4......
4.5.1Visual impairment questions asked......
4.5.2Visual impairment prevalence estimates......
4.5.3Methodological issues......
4.5.4Study contacts and further information......
4.6Family and Children Survey (FACs) 2006-2007......
4.6.1Visual impairment questions asked......
4.6.2Visual impairment prevalence estimates......
4.6.3Methodological issues......
4.6.4Study contact and further information......
4.7Family Resources Survey......
4.7.1Disability questions asked......
4.8FRS Disability Follow-up (1997)......
4.8.1Visual impairment questions asked in Disability Follow-Up......
4.8.2Visual impairment prevalence estimates......
4.8.3Methodological issues......
4.8.4Study contact and further information......
4.9Living in Wales Survey 2008......
4.9.1Visual impairment questions asked......
4.9.2Visual impairment prevalence estimates......
4.9.3Methodological issues......
4.9.4Study contact and further information......
4.10Low Income Diet and Nutrition Survey 2005......
4.10.1Visual impairment questions asked
4.10.2Visual impairment prevalence estimates......
4.10.3Methodological issues......
4.10.4Study contacts and further information......
4.11National Diet and Nutrition Survey (NDNS) 2008-9......
4.11.1Visual impairment questions asked......
4.11.2Visual impairment prevalence estimates......
4.11.3Methodological issues......
4.11.4Study contacts and further information......
4.12Life Opportunities Survey (LOS) 2010......
4.12.1Visual impairment questions asked......
4.12.2Visual impairment prevalence estimates......
4.12.3Methodological issues......
4.12.4Study contacts and further information......
4.13Understanding Society 2010/11......
4.13.1Visual impairment questions asked......
4.13.2Visual impairment prevalence estimates......
4.13.3Methodological issues......
4.13.4Study contacts and further information......
5Comparisons between surveys......
5.1Types of measures identified......
5.1.1Questions on disability and health conditions......
5.1.2Questions on visual capability......
5.1.3Questions on specific ocular conditions......
5.1.4Questions on self-rated eyesight and ‘difficulty seeing’......
5.2Inclusion and exclusion criteria......
5.2.1Use of low vision aids (LVAs)......
5.2.2Whether conditions are ‘limiting’......
6Conclusions......
Appendix ASummary of surveys......
Summary table of reviewed surveys, sample and questions......
Appendix BProject team......
Project team......
Project advisors......
Appendix CSuggested further work......
Summary......
C1Extend the scope of the review......
C2Cognitive development work with visually impaired people......
C4 and C5 Developing the HSE coding of visual impairment......
C6 and C7Secondary analyses of HSE and other survey datasets: trends, subgroups and social inclusion
Notes and References......
Tables
Table 4:1 Prevalence of visual impairment, by age group (HSE 2007)
Table 4:2 Prevalence of visual impairment, by sex (HSE 2007)
Table 4:3 Prevalence of visual impairment, by age group (SHeS 2008)
Table 4:4 Prevalence of visual impairment, by sex (SHeS 2008)
Table 4:5 Prevalence of self-reported difficulties in seeing the face of someone across a room, by age group (WHS)
Table 4:6 Prevalence of self-reported difficulties in seeing the face of someone across a room, by sex (WHS)
Table 4:7 Prevalence of self-reported visual impairment at Llticde, by age group (WHS)
Table 4:8 Prevalence of self-reported visual impairment at Llticde, by sex (WHS)
Table 4:9 Prevalence of self-reported cataracts/eyesight problems in past 12 months, by age group (APMS)
Table 4:10 Prevalence of self-reported cataracts/eyesight problems in past 12 months, by sex (APMS)
Table 4:9 Prevalence of self-reported ocular conditions, by age group (ELSA)
Table 4:12 Prevalence of self-reported ocular conditions, by sex (ELSA)
Table 4:13 Prevalence of self-rated quality of eyesight, by age group (ELSA)
Table 4:14 Prevalence of self-rated quality of eyesight, by sex (ELSA)
Table 4:15 Prevalence of self-rated difficulty in seeing, by age group (FACS)
Table 4:16 Prevalence of self-rated difficulty in seeing, by sex (FACS)
Table 4:15 Prevalence of ‘difficulties with sight’ i.e. uses glasses or contacts by age
Table 4:16 Prevalence of ‘difficulties with sight’ i.e. uses glasses or contacts by sex
Table 4:15 LOS % of respondents reporting difficulties in seeing things up close and at distances by age
Table 4:16 LOS % of respondents reporting difficulties in seeing things up close and at distances by sex
Table 4:15 Prevalence of visual impairment by age using LOS definition
Table 4:16 Prevalence of visual impairment by sex using LOS definition
Table 4:15 Prevalence of respondents reporting difficulties with sight by age group in Understanding Society
Table 4:16 Prevalence of Prevalence of respondents reporting difficulties with sight by sex in Understanding Society
Table 5:1 Visual impairment measures among older people, by survey and age group
Table 5:2 Visual impairment measures among younger people, by survey and age group
Figures
Figure 4:1 HSE 2007 question on long-standing illness, disability or infirmity
Figure 4:1 Long-standing illness code frame
Figure 4:3 Descriptions of codes related to visual impairment
Figure 4:4 SHeS 2008 ‘long-standing physical or mental condition or disability’ questions
Figure 4:3 WHS 2007 question on eyesight quality
Figure 4:4 WHS 2007 questions on health conditions and disabilities
Figure 4:5 Comparisons between ‘See’ and ‘LLticde’ by sex (WHS)
Figure 4:8 APMS 2007 Questions on general health
Figure 4:9 ELSA questions on ocular conditions
Figure 4:10 ELSA question on self-rated eyesight
Figure 4:11 Self-rated quality of eyesight, by whether have a self-reported ocular condition (ELSA)
Figure 4:12 FACS question on long-standing illness or disability
Figure 4:13 Comparing prevalence of visual impairments reported in HSE, SHeS and FACs
Figure 4:12 FRS questions on long-standing illness or disability
Figure 4:13 FRS Disability Follow-up questions on health conditions
Figure 4:13 FRS Disability Follow-up questions on visual capacity
Figure 4:14 Living in Wales questions on long-term illness or disability
Figure 4:15 Living in Wales questions on type of disability or impairment
Figure 4:16 LIDNS questions on long-standing illness, disability or infirmity
Figure 4:16 LOS questions on sight
Figure 4:16 LOS questions on long-term impairments and disability
Figure 4:17 Understanding Society questions
Figure 5:1 Phrasing of questions on disability
Figure 5:1 Prevalence of visual impairment, by survey and sex
Figure 5:2 Prevalence of visual impairment, by survey and age group
Acknowledgements
We would first like to thank the Thomas Pocklington Trust and its trustees for guidance on shaping this review and for the financial support that made it possible. In particular, we thank Dr Angela McCullagh, Katrina Harrington and Sarah Buchanan for their advice and patience.
We would also like to thank the many people who have contributed to this project in terms of identifying relevant resources for inclusion, providing detail on the methodology of specific surveys, and advising on the clinical aspects of conditions that can cause sight loss. We are particularly indebted to the staff at the UK Data Archive for help with accessing the datasets.
Furthermore, we would like to acknowledge the time and contribution of the many thousands of respondents whose generous participation continues to make social surveys possible, and the professionalism and dedication of the interviewers and the research nurses who collect the survey data in field.
Jo d’Ardenne, Matt Hall, Sally McManus
Executive summary [N1]
The Pocklington Trust
This review was funded by a research grant from the The Pocklington Trust. The Trust supports social and public health research that identifies practical ways in which the lives of people with sight loss can be improved. The report describes what national general population survey data are available that allow us to compare the situation of people with visual impairment with that of the rest pf the population. The objective is to improve the measurement of visual impairment in British national survey research.
Background
Registration rates and health service use statistics represent a minority of people affected by sight loss.Population prevalence estimates vary depending on the definitions applied and the data collection methodology used. A range of different indicators of visual impairment are used on national surveys in Britain. There have been no ‘gold standard’ questions for capturing the presences of visual impairment in general population survey research.
Aims
- Identify what questions aboutthe presence of visual impairment are currently being used in national surveys of the general population.
- Consider what these questions capture(for example, reference to low vision aids; use of filter questions; mode of administration; location in questionnaire; prevalence estimates of visual impairment generated; how rates compare across surveys by sex and age).
- Engage with a range of clinical and social researchers and with government social research and its agencies, on evaluating current approaches to assessing visual impairment on general population surveys.
Methodology
- A review was undertaken to identify relevant surveys using search engines and contacts. Inclusion criteria were applied to prioritise studies for the analysis.
- Datasets were obtained from the UK Data Archive or direct from researchers. For each survey, the sample profile was described and the questions and the format of their administration were documented.
- SPSS was used to manage the data: the relevant variables were identified, derived variables produced, and prevalence rates generated and presented by age and sex.
- The draft report was circulated to various experts and commissioners in the field for comment and a roundtable discussion held to refine the report’s content and recommendations.(underway)
Notes only:
Results
Discussion
Recommendations
- Harmonised disability questions group.
- Sample
- Survey topic
- Interviewer and research nurse training
- Survey method
- Question location – Health section? Disability section? Socio-demographics?
- Question format – prompted, unprompted, observed, objective.
- Filter questions
- Use of low vision aids, impact on functioning
- Code frames
- Aggregating categories: ‘sensory impairments’
- Survey reporting
- Archiving and documentation.
1Background and approach
1.1 Background
1.1.1Estimates of the prevalence of visual impairment
There are various estimates of the size of the population with visual impairment[i]: a widely cited figure is that sight loss affects the everyday lives of up to two million people in the UK.[ii] Access Economics estimated that in 2008 there were 1.8 million people in the UK with sight loss (1.6 million who were partially sighted and 0.2 million blind people).[iii]
Our changing demographics – including in terms of age, ethnicity and obesity – are impacting on trends in the prevalence of visual impairment.[iv]The majority of affected people are aged 75 and over,[v]although it has also been estimated that 80,000 people of working age are living with sight loss in the UK.[vi] The overall visual-impaired population has been projected to rise by a quarter over the next 30 years,[vii] and to more than double by 2050.[viii]In England, about 310,000 people are registered with Councils with Adult Social Services Responsibilities (CASSRs) as sight impaired (partially sighted) or as severely sight impaired (blind).[ix],[x] Registration rates therefore represent an underestimate of the wider number of people who may be considered to have a visual impairment.
1.1.2Factors influencing estimates of visual impairment
Prevalence estimates vary according to what is measured (for example, conditions or symptoms), how it is defined and captured, and who is asked. Visual impairment is not only multi-categorical in the symptoms manifested (for example, tunnel vision, blurred vision, reduced contrast), but also multi-dimensional in encompassing a spectrum of severity within each category. Self-evaluation of symptoms and severity are problematic, as they depend on expectations and recall as well as being sensitive to other sources of reporting bias.
While some thresholds of severity have been agreed for particular contexts (for example, to qualify for registration), there are no ‘gold standard’ questions for operationalising such thresholds in survey research. The work of Tait et al. has drawn together key studies that have contained good quality ophthalmic assessments, mostly among older people. However, most national, general population social surveys do not have the capacity (in financial or time resources) to include such assessments. Recently, the Office for National Statistics (ONS) issued a suite of harmonised questions recommended for the measurement of disability and impairment on government social surveys. These will be considered as part of this review.
1.1.3Use of general population surveys in estimating prevalence
To examine the prevalence of impaired vision, and its comorbidity and correlates, surveys are uniquely well placed. They collect a wealth of rich and publicly available data and are a cost effective resource that can be better utilised. In particular, general population surveys enable examination of treatment and registration rates among all people with visual impairment, as samplesalso include those who are not in contact with support and services. General population survey data are essential if the circumstances of visually impaired people are to be compared with those of the population as a whole. Without such data, social and economic inequalities can not be fully documented.
Major government and research council funded studies ask a range of core socio-demographic questions, irrespective of the focus of the questionnaire. These often use agreed government harmonised question wordings to enable cross study comparisons.[xi] General health, and to a lesser extent functional impairment, have been included in a number of these. However, visual impairment has not been prioritised. We can hypothesis why this might be:
- it is regarded as of specialist (rather than generalist) interest;
- it is not recognised as widespread;
- it is subject to a wider neglect of issues that disproportionately affect older people; and finally,
- because of the lack of good quality usable questions suitable for lay survey administration (see Tate et al.[xii]).
In the current economic climate, which has hit social research hard, these obstacles are even harder to overcome.
It should be noted that it is not an aim of this review to estimate the prevalence of visual impairment in the population, rather to identify what survey questions are currently in use and to examine the impact of question and survey method on the proportion and characteristics of the population with visual impairment that is identified.
1.1.4Variation in the definition of visual impairment
The current approaches to assessing whether or not survey respondents have a visual impairment are inconsistent and may often be inaccurate. Approaches in both clinical contexts and population-based studies have used varying methods and thresholds in defining and measuring sight related problems. Definitions have typically focused on specific medical eye conditions, the presence of particular symptoms, functional aspects of what the person can and cannot see in different contexts, and on subjective assessments of quality of eyesight. There is no ‘true’ or absolute prevalence of visual impairment. Rates and associations depend on the definitions used and in terms of what aspects of visual impairment are considered and what severity thresholds are considered relevant for being ‘visually impaired’.
1.2 Aims of this survey review
The aims of this report are to:
- Review what questions capturing any aspect of visual impairment are currently being used; with what populations; and where.
- Describe what these questions measure; what allowance is made for use of low vision aids; what prevalence estimates of visual impairment they generate; and how the rates compare by sex and age group.
- Engage with a range of clinical and social researchers and with government social research and its agencies on evaluating current approaches to assessing visual impairment on general population surveys.
We hope that through this review and subsequent discussions to:
- Review the data already collected.
- Raise the profile of visual impairment measurement among social researchers and survey commissioners to inform future data collection.
- Identify cost effective secondary analyses or methodological development work that could be done to situate the circumstances of visually impaired people in the context of the wider population.
1.3 Process of review
Stage 1: Initial consultation
- Initial exchange with experts (research, clinical and policy).
Stage 2: Review population based surveys including a measure of visual impairment
- Identify national general population surveys.
- Obtain data sets and generate prevalence estimates.
- Review methodology.
- Describe potential secondary analyses.
To come:
Stage 3: Discussions
- Present initial review findings.
Stage 4: Finalise review report
- Circulate for further iterations of input and comment.
1.4 Summary of methodological approach
A search of datasets held by the UK Data Archive was undertaken. Papers and reports accessed through Athens, Medline, and Social Care Online were scanned for mention of relevant British-based national surveys. We also used word-of-mouth suggestions from other researchers, in particular to identify surveys where the visual impairment data had not (yet) been written up in survey reports or as academic papers.
SPSS was used to generate prevalence estimates of different definitions of visual impairment from all the relevant national surveys for which we were able to obtain data. This used the relevant weighting variables, where these were archived on the dataset. The rates werenational (representing England, Scotland or Wales)and produced by age and sex, to facilitate comparison across surveys covering different age ranges.