Prevalence and Burden of Self-reported Blindness, Low Vision, and Visual Impairment in the French Community: A Nationwide Survey

Objective  To estimate the prevalence of self-reported visual impairment and its association with disabilities, handicaps, and socioeconomic consequences.

Methods  A national survey was conducted on a random stratified sample of 359 010 French citizens living in the community; 21 760 subjects were selected at random and 16 945 persons (78%) agreed to further questioning. Four thousand ninety-one randomly selected caregivers were interviewed. Four subgroups of subjects were defined (blind or light perception only, low vision or still have form perception, other visual problems, and no visual problems). These were compared after adjustment for age, comorbidity, and household size differences.

Results  The prevalence of blindness was 0.10% and of low vision, 1.94%. Subjects with blindness needed assistance with daily activities more often than subjects with no visual problems; they also needed more house modifications. Many subjects with blindness (46.8%) and subjects with low vision (29.0%) were registered for social allowances. Subjects with blindness had fewer paid activities (4.5%) than subjects with no visual problems (20.7%). Social allowances increased considerably (by €277) between those with low vision and those with blindness. Monthly household incomes were lower (P<.001) for subjects with low vision (€1255) and blindness (€1587) than for subjects with no visual problems (€1851).

Main Outcome Measures  Collected data included social demography, home description, household income, handicaps, disabilities, social allowances, and daily activities.

Conclusion  The results demonstrate associations between self-reported visual impairment and daily living.

Blindness is known to be one of the most severe disabilities to affect an individual, his or her family, and society.¹ In 2001, according to the World Health Organization (WHO),² visual impairment was responsible for 2 286 000 disability-adjusted life-years in the high-income countries.

The use of registers to estimate the prevalence of blindness is controversial since a high proportion of subjects with blindness do not register.^3-6 Other surveys estimate prevalence only in the very elderly population.^7,8 According to a review of the available prevalence data on blindness performed by the WHO,⁹ using data up to 1994, 10 surveys have been conducted in Europe. Three of these were based on registers. A recent update was performed in 2002.¹⁰ The prevalence of low vision (LV) was estimated by only 1 survey (2.0%), which was in Ireland.¹¹ Prevalence of blindness in the European Union countries ranged from 0.09% to 2.6%, depending on the target population. The authors commented on a lack of national data for most European countries.

The social consequences of blindness (disability, dependency, and decreased household income) have rarely been evaluated in representative samples of national populations. Some surveys were conducted to evaluate the impact of blindness on daily activities.^12-16 Most of these reports were concerned with survey methods. Quality of life, as a global measure of the consequences of handicap, was more often studied.^17-25

The economic consequences of blindness have also rarely been documented. The global cost of blindness was estimated as a quarter of the United Kingdom gross national product.²⁶ The cost of blindness to the Australian government and community was estimated as somewhere between $9749 (Australian dollars) and $26 720 (Australian dollars) per patient per year.²⁷ The economic burden of blindness in India was estimated as $4.4 billion (US dollars) for 1997.²⁸ Lastly, the mean cost per person with blindness per year was $11 896 (US dollars) in 1990 in the United States and totaled $4 billion (US dollars).²⁹

The present survey had 3 aims¹: to estimate the prevalence of self-reported blindness and low vision in French subjects living in the community, on a national basis²; to study the consequences of self-reported blindness and LV, focusing specifically on disabilities (restricted ability or inability to perform daily activities) and handicaps (restricted ability or inability to fulfill desired social roles)³; and to collect information on the economic consequences of visual impairment on income and social allowances.

This survey was conducted at the request of the French State, which also provided funding. The data were gathered by Institut National de la Statistique et des Etudes Economiques, which is the state agency responsible for executing regular national census surveys. This survey complied with all existing national regulations, including personal data privacy and access. The database was subsequently made available to researchers for secondary analyses.

A national census survey is performed almost every 10 years in France. Data collected include demographics, employment, housing, teaching, nationality, and immigration but no information on handicaps, incapacity, and dependency. France is divided into 26 regions, 100 counties, 4032 districts, and 36 679 cities. Groups of interviewers are responsible for a specific geographical area that respects the region, county, district, and city boundaries. Each household is visited by an interviewer belonging to a geographical area. Data are collected on each member of the family. All French citizens are questioned, and answering to the national census is compulsory according to French law. Questionnaires are completed with or without the help of the interviewers but a final check is always conducted.

The aim of the Handicap Incapacity Dependency (HID) survey was to document handicaps, incapacity, and dependency at a national level, with no limitation on age, in French citizens living in the community (citizens living in long-term institutions were studied in another survey). It was decided to piggyback on the 1999 national census survey to decrease data collection costs and to get 1999-updated weights that could be estimated at a national level.

The design of this survey followed the guidelines and principles for developing disability statistics published by the United Nations.³⁰ The sample was selected (Figure) according to a 2-step process.^30,31

• The last national census survey was performed in 1990 (Figure); 57 831 816 citizens were documented and statistics on geographical area (see earlier definition) were available. Two thousand two hundred seventy-five geographical areas were picked at random from the 1990 survey (since the 1999 survey results were not available), stratified by counties nested within regions and family and socioprofessional statistics. Interviewers were picked at random in each geographical area. During the 1999 national census, a filtering survey called Everyday Life and Health (ELH) was added. It consisted of a self-administered, 18-item questionnaire that collected information on daily activities. In total, 2223 (97.7%) of the 2275 geographical areas collaborated in the ELH survey. From the 399 784 questionnaires distributed, 359 010 (89.1%) were completed, returned, and checked by the interviewers. Questionnaires were to be answered by all members of a household. Weights for extrapolating the data to the entire population were estimated from the 1999 national census.³²

• The HID survey was performed to quantify handicaps and dependencies. Subjects from the ELH survey were divided into 6 handicap groups ranging from no handicap (group 1) to severe handicap (group 6), based on a handicap severity score.³² Subjects in the severe handicap group had a higher probability of being detected by the HID survey than did those in the ELH survey (eg, patients in the most severe group represented 30.9% of the HID population but only 4.0% of the ELH population). This oversampling method made it possible to describe the consequences of handicaps in detail since citizens with handicaps were overrepresented in the HID survey. Face-to-face interviews were available for 16 945 (77.8%) of the 21 760 subjects selected at random from the ELH respondents. During these interviews, if a caregiver was present, he or she was asked to answer questions about their own lives and freedom. In total, 4091 (97.1%) of 4969 domestic caregivers agreed to answer a caregiver-specific questionnaire. Weights for national extrapolation were calculated on the basis of the ELH handicap severity classification, refusal to participate in the HID survey, age, sex, and size of household, type of household, and geographical area size.

Both questionnaires are available at the Institut National de la Statistique et des Etudes Economiques on request by researchers.

In the ELH survey, demographics, data on near and far vision, speaking ability, hearing, daily activities, need for assistance or devices, and reasons for the handicaps were collected. Presence or absence of the handicap was self-reported by each member of the household on a 3-page questionnaire.

In the HID survey, a trained interviewer filled in the questionnaire during face-to-face interviews at home using remote data-entry computer software. The data (up to 800 variables per questionnaire) included type of handicap and disabilities (if any), indexes of daily activity (Katz, Colvez, and Etablissements d’Hébergement pour Personnes Âgées indexes^33,34), the reasons for and duration of handicaps, social professional and family environment, caregivers’ and subjects’ social demography, household characteristics, home modifications required for the handicap, mobility difficulties, household income (including social allowances), social allowances, public and private health insurance, copayment, and current leisure activities. The mean interview duration was 40.6 minutes for subjects and 3.5 minutes for caregivers.

The presence of handicap was identified by the following initial yes/no question: “In everyday life, are you faced with either physical, sensorial, intellectual, or mental difficulties (resulting from an accident, a chronic disease, a problem at birth, a disability, aging . . . )?” Independent of the previous answer, the following question was asked, aimed at listing handicaps: “What kind of difficulties, disabilities, or other health problems do you suffer from?” All the answers were written down without any alteration. If during the rest of the questionnaire the interviewers came across diseases or other health problems not mentioned earlier, they were also reported. The declared morbidity was then medically coded by experts in this field, using International Classification of Diseases (ICD) classification.

For the purpose of the present analysis, the presence of blindness or LV was accepted if declared by the subject and not medically assessed. During the ELH survey, subjects answered the following 3 questions: (1) Do you have trouble reading the plain characters in a newspaper article (using spectacles or contact lenses if you normally wear them)? (2) Do you have trouble recognizing the features of someone across the room or street (using spectacles or contact lenses if you normally wear them)? (3) Do you usually have trouble filling out a check or simple form on your own? Subjects answered the following questions when selected for the HID survey: (1) What is the medical name of the disease you are suffering from, if you know it? (2) Are you completely blind (or light perceptive only)? (3) Would you say you are completely blind (light perception at best), partially blind (still have form perception), or visually impaired? (4) Are you suffering from other visual problems (visual field deterioration, color perception, strabismus, etc . . . )? Question 1 (near vision) and question 2 (distant vision) were repeated in the HID survey. Only subjects who declared vision problems (identified during ELH/HID interviews) had to answer the 4 HID survey questions. Subjects were classified into 1 of the following groups on the basis of their answers: (1) blind; (2) LV; (3) other visual problems (OVP); or (4) no visual problems (NVP). Those with blindness declared either light perception only or total loss of vision. Those with low visual acuity declared either severe difficulties in long- or short-distance vision (questions 1 and 2, ELH and HID surveys) and in form perception or visual impairment (question 3, HID survey). Subjects who declared visual troubles but were neither blind nor had LV were classified as having OVP. Therefore, a person with OVP was someone with an eye-related condition not sufficiently severe to qualify as LV or blindness. The OVP group was included to ensure that LV denoted a distinct subgroup of visual impairment.

Social allowances covered the following groups of items: allowance for adults with disabilities, compensatory allowance, income guarantee, special education allowance, housing allowance, special dependency allowance, disablement paid by the State, disablement allowance deriving from an accident at work, daily allowance paid by the French Sick Fund, allowance paid by an insurance company, military disablement allowance, and other. Details within groups were collected. Results were expressed in euros, €1 fluctuating between $0.85 (US dollars) and $1.15 (US dollars).

Disabilities were described according to an ordinal scale following a question starting with “Can you . . . ?” Answers varied according to the handicap, but the general structure was as follows: “Irrelevant,” “Yes, without assistance and without any trouble,” “Yes, without assistance but with some difficulty,” “Yes, without assistance but with much difficulty because of my physical disorders,” “No, I need partial assistance,” “No, I need assistance for everything,” and “Will not answer or does not know.” The data presented in the tables grouped partial assistance and assistance for everything together.

All analyses were conducted with SAS software, release 8.2 (SAS Institute, Cary, NC). Since the 4 subgroups were not comparable, adjustments were made using a weighted logistic regression for qualitative variables and a weighted analysis of variance for quantitative parameters.

Adjustments were performed on the covariables age, number of comorbidities, and number of subjects in a household, since these variables were found to be linked to almost all of the outcome parameters. Rates were adjusted by using the blind population estimates as reference; estimators of the logistic regression were applied to the covariables estimated on the blind population.

Odds ratios were calculated by using the population without visual problems as a reference.

All tests were interpreted 2-sided, and α was fixed at .05. No corrections were applied to take account of test multiplicity.

The response rates were 89.8% and 77.8%, respectively, in the ELH and HID surveys. Demographics of nonresponders were not collected. In the HID survey, 12% of the households refused to answer, 7.3% of households were not accessible (interviews were conducted just after the great French storm of 1999), 0.7% of the subjects had died since the 1999 national census, and 0.5% were living in institutions.

Of 16 945 subjects interviewed, 87 were classified as blind, 1126 had LV, and 1061 claimed a visual problem. On extrapolation to the whole community population (citizens living in long-term institutions were not included), the corresponding prevalence figures were 57 959 individuals with blindness (0.10%), 1 116 862 with LV (1.94%), and 1 672 111 with OVP (Table 1).

The mean age of subjects with blindness exceeded that of persons with NVP, subjects with partial sight, and those with OVP by 35.3, 24.4, and 14.2 years, respectively. Subjects with impaired vision also declared more handicaps, and the differences persisted after age adjustment (P<.001). Subjects with blindness and LV had more frequent motor and auditory handicaps than those with NVP. Overall, an estimated 1.588 million individuals had difficulties in performing 1 of the following daily activities without assistance³⁵: washing, dressing, going to bed, getting out of bed, sphincter control, and eating prepared food (Katz index). After age adjustment, 7.7% of subjects with NVP, 10.4% of those with LV, and 37.6% of subjects with blindness needed help performing the activities cited (P<.001).

Table 2 describes the help required to perform most daily activities by vision status after adjustment for age, number of comorbidities, and size of household. Overall, subjects with blindness had greater difficulty in performing daily activities than those with NVP. A significant difference also existed between subjects with LV and subjects with blindness. More than two thirds of subjects with blindness could not undertake the following 7 activities on their own: shopping, paperwork, travel, housework, preparing a meal, using an elevator, and taking medicine.

Consequently, a caregiver often had to support a person with blindness or LV. Subjects with blindness demanded significantly more of a caregiver than did subjects with NVP. A similar disparity was observed between subjects with LV and subjects with blindness. More than a quarter of the caregivers could not leave their blind charge for a whole day, and more than half declared that their physical and mental welfare was affected. Many of them (56.1%) had to reorganize their professional activities (Table 3).

Table 4 describes reliance on paid services in relation to the severity of visual impairment after adjustment for age, number of subjects in the household, and number of comorbidities. Subjects with blindness more frequently needed surveillance, nursing, and other paid services than did subjects with LV or NVP. The latter 2 groups were similar to each other.

After age adjustment, subjects with blindness declared a more frequent need for specifically adapted household items (ie, tables, chairs, beds, other pieces of furniture, ramps, and kitchens) than did other groups (Table 5). Individuals with LV expressed a need for lavatory/toilet and bathroom modifications. A change of residence, necessitated by the visual impairment, was 5 times more frequent among subjects with blindness (15.6%) than subjects with NVP (3.6%). A further 9.7% of subjects with blindness envisaged a future move because of their visual condition (3 times greater than foreseen by subjects with NVP [3.3%]).

The use of most devices and other aids specific to blindness increased as vision decreased (Table 6). The need for sticks, white sticks, walking aids, wheelchairs, and dogs was well met. However, a tremendous unmet need was identified for optical assistance, computer interfaces, software adaptations, and tape recorders.

Individuals with NVP were 5 times more likely than persons with blindness and nearly twice as likely as those with LV to have a paid professional activity (Table 7). More subjects with blindness (44.1%) than subjects with LV (12.7%) or NVP (7.4%) received social allowances. Monthly average allowances to subjects with blindness were markedly greater than those provided for subjects with LV (€364 vs €87). However, total incomes were only €62 more for blind as compared with LV households. Lastly, the total income (including social allowances) of households with persons with blindness and partial sight was lower, by more than €300, than for households with no visual difficulties. On average, social allowances represented less than 16% of the income lost because of LV.

The design of this survey followed the guidelines and principles for developing disability statistics published by the United Nations.³⁰ Instead of classifying patients according to their disease status using the ICD,³⁶ this survey tried to measure the consequences of handicaps and to reproduce in some way the Wood sequence³⁷ (ie, impairments, disabilities, and handicaps). The 2-step sampling method allowed us to oversample patients with handicaps and thereby led to more accurate estimates. The response rates of both surveys were higher than 75%, and reasons for nonparticipation were documented. Weights based on the demographic features of subjects who refused to participate permitted bias to be minimized during the nationwide extrapolation of the data to the whole French population living in the community. However, the fact that 25% of the citizens did not respond is a limitation of this study.

The cross-sectional design of this survey did not permit an analysis of possible causal relationships between blindness, handicap, dependency, and incapacity. Follow-up data collection should be performed to confirm our findings.

The visual acuity of responders was neither measured nor controlled by an ophthalmologist or an optometrist and we were unable to apply the WHO classification for blindness (visual acuity<3/60; ICD-10). This issue was already discussed in a similar survey conducted in institutions.³⁸ Subjects classified with blindness in this survey were those who self-declared that they did not perceive shapes. The categorization of subjects with LV was much more difficult since some who declared they had only shape perception might have been classified with blindness according to the WHO. However, we believe that the interviewer-administered questionnaires were of sufficient quality to correctly differentiate between LV and blindness with a high probability of accuracy. Indeed, in this elderly population, chronic diseases such as age-related macular degeneration and glaucoma are known to account for the vast majority of ocular disorders. Because of the chronic nature of these conditions, we estimated that patients had become educated at accurately assessing their visual handicap.

Our self-reported blindness prevalence estimates are very close to the low figures published for Western European countries. The prevalence of blindness ranges from 0.09% to 2.6% according to Thylefors et al⁹; our estimate was 0.1%. However, of the 7 surveys based on random samples, only 3 were performed on a nationwide scale (Italy, Malta, and the United Kingdom). The sole French survey, conducted in the city of Marseilles (1985), found a prevalence of 0.2%.³⁹ Also, the prevalence of LV (1.94%) in our survey was very close to that in Ireland (2.0%), measured by the 1 survey available.¹¹ Lastly, our figures are probably conservative since patients in institutions were not taken into account. It is likely that the probability of living in an institution increases dramatically with blindness.

People with NVP reported less comorbidity than subjects with visual impairment, even after adjustment for age. This might be explained by a significant number of subjects (with blindness, LV, or OVP) who have systemic diseases, which can lead to ocular and other disorders. Eye conditions predisposing to loss of vision must be diagnosed and treated together with such systemic diseases.

In general, some associations were found between vision level of handicap and disabilities. Subjects with partial sight needed more daily care, either from paid help or care provided by members of the household. The disparity between subjects with LV and blindness was far greater.

Also, some relationship between visual handicaps and home modifications and the need for some medical devices was found. Families with a member with blindness moved 5 times more frequently than households with a member with NVP. Since home modifications and moving are not subsidized, many subjects must cope in unsuitable premises. Traditional devices and aids for blindness are reimbursed by French Sickness Funds, but a tremendous need exists for computer aids. The utility and cost-effectiveness of such aids should be further evaluated for possible reimbursement.

Only 44.1% of people with blindness living in the community registered for a social allowance. This confirmed previous conclusions that the prevalence of blindness cannot be extrapolated from social allowance registers.^3-6 Efforts to inform subjects with blindness of their social rights should be reinforced.

People with blindness engaged in a paid professional activity far less frequently than other groups in the survey. Social allowances increased steeply with the transition from LV to blindness. The allowances composed a much higher proportion of total income in blind households, whereas total income was only slightly greater in blind than LV households.

This quantification of disabilities associated with self-reported blindness is very important for public health decision makers. It is very important to know whether treatments related to vision are cost-effective from both a resource allocation and a between-patient equity point of view. Aging by itself is also a source of disability. This study brings nationwide estimates (rates and odds ratios) of the disabilities associated with self-reported blindness adjusted for age (since people with blindness are older). Based on these data, a specific nationwide cost of blindness could be calculated and decisions could be made on how much should be budgeted to provide fair indemnification to citizens with blindness and LV. It could also serve to evaluate the added value of innovation that would avoid or postpone LV and/or blindness in subjects living at home.

In conclusion, this survey suggests that medical procedures (eg, prevention, diagnosis, and treatment) aimed at preserving shape perception might help people at risk of blindness to maintain their independence and minimize the social impact of progressive visual impairment on their caregivers. These data will be critical to quantify the return on investment from any innovations that will contribute significantly to maintain low visual function. Prospective subject follow-up will be needed to confirm these findings.

Correspondence: Antoine Pierre Brézin, MD, PhD, Service d'ophtalmologie, Université Paris-Descartes, Faculté de Médicine, Hôpital Cochin, Université Paris 5, 27 rue du Faubourg Saint-Jacques, 75679 Paris CEDEX 14, France (antoine.brezin@cch.ap-hop-paris.fr).

Submitted for Publication: October 22, 2003; final revision received March 14, 2005; accepted December 22, 2004.

Financial Disclosure: Dr Berdeaux was employed by Alcon Research Ltd, Alcon France SA, Rueil-Malmaison, but the authors have no financial or other conflicts of interest relating to Alcon Laboratories, Fort Worth, Tex, or directly relevant to the contents of this study. Dr Brézin has participated in clinical trials, unrelated to this study, financed by Alcon Laboratories.

Funding/Support: This study was supported by an unrestricted grant from Alcon France SA, Rueil-Malmaison, France.

Additional Information: This study was conducted in collaboration with Cemka-Eval SA, Bourg-la-Reine, France, and according to local laws.

Acknowledgment: We are grateful to Irène Fournier, PhD, and Benoit Riandey, PhD, for the database.