Recherches récentes Mes recherches
Vous consultezDid the Prevalence of Disability in France Really Fall Sharply in the 1990s?A Discussion of Questions asked in the French Health Survey
AuteursEmmanuelle Cambois[*] [*] Institut national d’études démographiques, Paris. ...Emmanuelle Cambois, Institut National d’Études Démographiques, 133 boulevard Davout, 75980 Paris Cedex 20, France, phone: 33 (0)1 56 06 22 55, e-mail: firstname.lastname@example.org
suitedu même auteur
For nearly thirty years now, statistical surveys on the health status of the general population have extended in scope beyond disease and medical consumption to include measures of disability. At a time of general mortality decline when people can expect to live for longer with illness and disability, their goal is to evaluate quality of life, autonomy, and social participation. Initially focused on seniors’ health, disability questions now cover the population as whole. They are ever more numerous and detailed, providing a fuller description of disabilities and of respondents’ reliance on assistive devices or assistance from their contact circle so as to improve understanding of current and future needs in this area.
2 Recent international data broadly indicate a decrease in disability in the 1990s, concurrent with a rise in life expectancy (Robine et al., 1999; Robine et al., 2003; Crimmins, 2004). For example, the prevalence of disability at all levels has been declining steadily among U.S. seniors (Freedman et al., 2002), though more slowly at the oldest ages (Schoeni et al., 2005). It is accompanied by a decrease in reliance on personal care assistance among people aged 70 or over (Freedman et al., 2004). Studies show a comparable fall in disability in Europe as well, for example in Sweden (Lagergren and Batljan, 2000) and Denmark (Bronnum-Hansen, 2005).
3 In France, a general question on disability – more specifically, on being hampered or disabled – was introduced in the ten-year Health Survey of 1980-1981 and repeated in 1991-1992. Analyses have shown a decrease in disability in the 1980s, both at all levels and at severe levels only (Robine and Mormiche, 1994). In the same interval, findings show an increase in reporting of certain illnesses but a loosening of their link to disability (Robine et al., 1998). The being hampered or disabled question was again included in the 2002-2003 survey, so it should be possible to track ongoing change. The raw results suggest a sharp drop in disability between 1991-1992 and 2002-2003 which, though expected, exceeds the most optimistic forecasts. To explain this finding, we explored the factors that may have contributed to this trend.
4 The substantial modifications in the 2002-2003 survey protocol may have “changed the nature” of the being hampered or disabled question and exaggerated the decline in measured disability. The question was moved to a different position, so its environment and the guidelines for responding to it were changed. Given the question’s subject and wording, it may have been sensitive to the new survey design. Our paper uses various data sources and disability indicators to discuss this hypothesis.
5 Part one describes the context of our study and the data. Part two gives a detailed presentation of the results obtained with the being hampered or disabled question in the three Health Surveys and the sharp decrease observed in the 1990s. In part three, postulating a change in the way the being hampered or disabled question was answered, and, consequently, in the characteristics of the population that it identifies, we describe the population using the other disability indicators available in the 2002-2003 survey. We conclude by offering some possible explanations for the steep fall in reported disability between the 1991-1992 and 2002-2003 surveys.
6 The universe of disability is complex. Broadly speaking, it encompasses the consequences of illness and accidents on people’s daily life and on their social integration. “Disability” is a generic terms that cover different dimensions such as self-perceived disability, functional health, need for help in everyday activities, and administrative recognition. These dimensions overlap only in part (Ravaud et al., 1999; Ravaud et al., 2002). Each describes a specific aspect of disability; some from a medical angle (such as impairments and functional limitations), others from a more social angle (such as administrative recognition, insurance coverage and social participation). Disability was first measured in general population surveys in the 1970s. The medical dimension of disability was addressed by studying functional, sensory, motor, and cognitive problems as well as restrictions in essential everyday activities (Robine et al., 1997).
7 In France, the questionnaires of the ten-year surveys on health and medical care (Enquête sur la Santé et les Soins Médicaux, ESSM) have included the being hampered or disabled question since 1980-1981 and have gradually added other instruments for measuring disabilities. In the late 1990s, a general population disability and dependence survey (Enquête Handicap, Incapacités, Dépendance, HID) made it possible to explore more aspects of disability (Mormiche et al., 2000). The 2002-2003 Health Survey questionnaire was expanded to take account of HID findings.
8 The ten-year ESSM surveys cover the private-household population. They seek to assess the population’s health status but also visits to the doctor and consumption (visits to the doctor, examinations, treatments, and medication) and are based on a specific methodology. In addition to the questionnaire on health problems and behaviours, they are based on a record of all visits to the doctor, prescriptions and health expenditures in a given period. An interviewer visits the household several times to record consumption items, and how and why they were consumed.
9 The 1980-1981 and 1991-1992 surveys were based on a household approach, i.e. one household member was in charge of responding to the questionnaire, on his/her own behalf and for each of the other household members. The “main respondent” would meet the interviewer to fill in the questionnaire. The other household members did not need to be present, as information concerning them was provided by the main respondent. As with individual data, the survey data were analysed independently for each household member according to his/her individual characteristics, recorded in the questionnaire. The questionnaire identified the main respondent so that any proxy effect on responses concerning another household member could be detected.
10 In the 1980-1981 and 1991-1992 surveys, information was collected over a three-month period, during which the interviewer would visit households five times. The being hampered or disabled question was asked on the fifth and final visit. As with the other questions, the responses could be cross-tabulated individually against socio-demographic characteristics and health information. The question was general, so as to cover a wide range of situations of varying severity. The wording mentioned both “being hampered or having difficulties in daily life” and “being disabled”:
“Yes” / “No”
“If yes, please specify the person”.
12 The Health Survey methodology was substantially revised in 2002-2003 (Appendix 1). In addition to fewer interviewer’s visits and a shorter collection period (three visits in a two-month period), the survey switched from household questions to individual questions in order to obtain more precise data on health status, medical consumption and preventive behaviour. In the new survey, each household member answered all the questions on his/her own behalf. Only an introductory section was filled in by a person who answered on behalf of all household members. This section mostly includes socio-demographic information but also some questions on health, including the being hampered or disabled question. This question was moved up to the start of the questionnaire so as to retain the household collection method that was used in the 1980-1981 and 1991-1992 surveys. The wording of the being hampered or disabled question was unchanged from previous waves. However, it was now asked during the interviewer’s first visit after a series of questions on standard socio-demographic characteristics. Its new context is thus more remote from the healthcare and health-problem issues that surrounded it in earlier surveys, where it came at the end of the questionnaire. It was not deemed necessary to determine who completes this household section, since the questionnaire core consists of individual questions.
13 During the fifth visit of the 1991-1992 Health Survey, the interviewer administered a specific questionnaire to a household member chosen at random (main respondent or another member). This individual interview made it possible to address personal topics for which a third party would find it difficult to provide answers, such as self-perceived health and health-related behaviour. If the person picked at random was over 65, a two-part module of questions on functional health was administered. The first part covered physical and sensorial functional limitations (locomotion, joint mobility and object grasping, hearing, eyesight); the second part concerned restrictions in domestic and personal care activities.
14 Drawing lessons from the HID survey and other countries’ projects in the area, the 2002-2003 survey assigned a prominent position to disability. The questions we used are given in Appendix 2. They comprise (i) general questions on disability: in addition to the being hampered or disabled question, we included one on long-term activity limitations (have you been limited by health problems for at least six months in activities people usually do?) and one on self-perceived disability (do you regard yourself as disabled?); (ii) a module of questions – for all adult household members – on physical and sensorial functional limitations and on restrictions in different activities; (iii) questions on restrictions in occupational activity (for health reasons, economically inactive and/or works part-time and/or has changed jobs and/or has interrupted his/her activity for more than six months); (iv) questions on chronic illnesses, included in this study as possible sources of disability (has a chronic or long-term illness or health problem; has total insurance coverage for long-term illness); (v) a question on the administrative recognition of a disability (do you have an officially recognized disability?). From these questions, we developed an aggregate functional health indicator, based on the nesting effect of disability situations discussed in the literature (Box 1) to allow a more precise description of individual disability situations (Guralnik et al., 2003; Simonsick et al., 2001). This indicator will enable us to identify the nature of the problems that prompt respondents to report disabilities via the being hampered or disabled question and the other two general questions (self-perceived disability, activity limitations).
Functional limitations and activity restrictions
So-called “biomedical” models refer to a process of loss of autonomy due to illness and health problems (Wood and Badley, 1978; WHO, 1980; Verbrugge and Jette, 1994; Fougeyrollas, 1995). In these models, difficulties often arise in a progressive and cumulative sequence: first, functional limitations (difficulties in near or distant vision, hearing, walking, etc.) for which the person manages to compensate; then functional limitations that persist, despite the use of assistive devices and that, in some cases, entail difficulties in domestic activities (shopping, housecleaning, everyday chores, etc.). In the most severe situations, people experience restrictions not only in domestic activities but also in personal care (washing, dressing, etc.) (Jagger et al., 2001; Barberger-Gateau et al., 2000). These levels of difficulty are generally interconnected, although some people experience activity restrictions without functional limitations (Cambois et al., 2005): 82% of respondents with restrictions in personal care activities reported problems with domestic or work activities and 98% have sensorial or physical functional limitations; 80% of persons with restrictions in domestic or work activities (without restrictions in personal care activities) reported sensorial or physical functional limitations.
On the basis of these models, we developed an aggregate functional health indicator divided into 5 mutually exclusive levels measuring different situations. Each level aggregates the responses to questions on chronic illnesses, functional limitations and activity restrictions: (1) no illness, functional limitation, or activity restriction; (2) chronic illness or functional limitation (eyesight, hearing, locomotion, etc.) compensated by an assistive device but without impact on activities (no activity restriction); (3) at least one functional limitation not compensated by an assistive device but no activity restriction; (4) at least one restriction in domestic or work activities but no restriction in personal care; (5) at least one restriction in personal care activities.
15 The study population comprises persons who answered the questionnaires during the interviewer’s three visits and who answered questions in the modules on functional limitations and activity restrictions; it comprises around 25,000 men and women aged 20 or over (of which 5,000 aged 65 or over) (Box 2)  The modules on functional limitations and activity restrictions...
Sample of 2002-2003 Health Survey respondents and study population
For the 2002-2003 Health Survey, we drew 25,000 address records at random, of which 86% were in the scope of coverage (occupied main residences); 8% of households refused to participate, 7% could not be contacted and 4% could not be surveyed. The sample contacted by the interviewers for the first visit includes 40,832 people living in private households in metropolitan France; it represents 78% of households chosen at random and in the scope of coverage. Of this group, 39,903 responded to the entire first-visit questionnaire. Sample attrition was larger between the first and third visits than in the 1991-1992 wave. The interviewers ultimately obtained 35,080 completed questionnaires. A weighting system adjusts the third-visit sample according to the characteristics of drop-outs. The section on functional health problems covered in the third visit concerns only individuals of legal age who, in the first visit, were found eligible to respond to the survey. The analyses using third-visit variables cover 26,126 persons of legal age (about 300 persons were disqualified, or 1.2% of adults in the first-visit sample). In this study, we work on the entire population, when the data permit, or on the adult population when using data from the third visit. To simplify the analyses, we work on five-year age groups and report results for people aged 20 or over, with special attention to the oldest group (65 or over).
16 In 2002-2003, 6% of persons living in private households, irrespective of age, were identified as having being hampered or disabled (7.6% of the 20+ age group). The prevalence is higher for women, at 6.3% versus 5.7% for men (respectively 8.0% and 7.2% among people aged 20 or over). Prevalence rises sharply with age, reaching nearly 35% for men and 41% for women aged 90 or over (Figure 1). The study of age-specific prevalence shows minimal gender differences. Indeed, a standardization of women’s ages brings the prevalences among both genders to the same level, for the total population as well as for the over-65s.
Prevalence of being hampered or disabled by five-year age group, 2002-2003
17 Earlier surveys found far higher age-specific prevalences of being hampered or disabled, with wider gender differences (Figure 2). These figures suggest a substantial decrease in age-specific prevalence rates for being hampered or disabled between 1991-1992 and 2002-2003, in particular among the oldest old; the decrease is far steeper than that observed between 1980-1981 and 1991-1992. We estimated annual rates of change for age-specific prevalence and total prevalence from the parameters of a log-linear regression. In the 1980s, the age-specific rates were stagnant or mildly increasing among the under-50s, an age group in which prevalence is low. After 50, the decrease fluctuated, according to the age group, between 0.1% and 4.3% per year among men, and between 0.9% and 3.5% among women. The 2002-2003 survey data point to a decrease of as much as 7% per year for all age groups in the 1990s. Overall, we estimate that the annual change in prevalence among people aged 65 or over (standardized on the population’s age structure in 2003) came to –5.5% between 1991-1992 and 2002-2003, compared with –1.6% between 1980-1981 and 1991-1992.
Proportions of men and women with disabilities (being hampered or disabled) by five-year age group since early 1980s
18 This decrease in the prevalence of being hampered or disabled also differs considerably from the patterns shown by international data. Applying the same model as above to data published in North American studies, we find that the use of a helper or an assistive device for personal care activities decreased by an average 0.7% per year in the 1990s among people aged 70 or over. The decline is steeper, but consistently under 2%, for the severest difficulties involving assistance (between –1.4% and –1.7%, depending on the assumptions). We find that even the most severe levels of disability – which appear to register the largest decreases according to the various studies – do not post variations as wide as the decrease measured by the being hampered or disabled question between 1991-1992 and 2002-2003.
19 The French section of the European Community Household Panel (ECHP) set up by Eurostat (1995-2001) contains a general question, “Are you hampered in your daily activities (occupational, domestic, recreational) by a chronic illness or a disability?”, placed in the middle of a questionnaire on general living conditions. This question identifies a far larger share of the population than the being hampered or disabled question: 25% of people aged 20 or over and 55% of people aged 65 or over in France in 2001. For the period 1995-2001, the standardized prevalence registered a mild annual change (up 0.2% among people aged 20 or over and down 0.02% among people aged 65 or over). Even though this prevalence is not comparable with that of being hampered or disabled, it does not suggest a major decrease that could in any way corroborate the massive drop measured by the French Health Survey.
20 These results prompted us to explore in greater detail the responses to the being hampered or disabled question in 2002-2003 and to suggest possible explanations for the decrease.
21 Of the three general questions in the 2002-2003 Health Survey, the one on being hampered or disabled – specifically administered at the household level – represents the most restrictive approach, despite a wording that refers to “simply being hampered or having difficulties” (Table 1): 7% of men and 8% of women aged 20 or over are identified (and respectively 15% and 17% of the over-65s). The question on long-term activity limitations (“limited by health problems for at least six months”), asked shortly afterwards in the individual questionnaire, selects nearly twice those percentages (14% of men and 16% of women aged 20+; 29% and 33% respectively of the over-65s). The prevalence of long-term activity limitations remains significantly higher for women than for men after age adjustment, contrary to the results of the being hampered or disabled question. The “Do you regard yourself as having a disability?” question, asked of individual respondents at the end of the survey, identifies almost 15% of men and women aged 20 or over and 30% of the over-65s.
Table 1 - Respondents reporting being hampered or disabled, long-term activity limitations, and self-perceived disabilities in 2002-2003
22 The three general questions overlap only in part, as they represent different approaches to disability. The majority of people identified as being hampered or disabled also report long-term activity limitations; likewise, most regard themselves as having a disability (in both cases, two-thirds of people aged 20 or over and three-quarters of the over-65s). By contrast, many respondents were not identified by the being hampered or disabled question, despite their reporting long-term activity limitations or self-perceived disabilities. This prompts us to examine the kinds of difficulties and disability situations that respondents reported in their answers to the being hampered or disabled question in 2002-2003.
23 A simple cross-tabulation with the aggregate functional health indicator shows that overall, the being hampered or disabled question tends to cover activity restrictions rather than functional limitations alone. In fact, 66% of men and 72% of women aged 20 or over who are hampered or disabled report restrictions in domestic or work activities or in personal care activities, which correspond to levels 4 and 5 of our indicator (Table 2). And among those who report no activity restrictions, nearly all at least mention residual functional, physical or sensory limitations, i.e., not compensated by assistive devices (17% of men and 12% of women aged 20 or over identified by the being hampered or disabled question) or compensated chronic or functional problems (12%). Ultimately, fewer than 5% of persons identified by the household question as being hampered or disabled do not report any of the health problems measured by the aggregate functional health indicator (and none in the 65 or over age group).
Table 2 - Breakdown of responses to general questions on being hampered or disabled, long-term activity limitations and self-perceived disabilities by functional health indicator level (%)
24 By comparison, we find that a majority of the answers to the general question on long-term activity limitations (posed individually) concern activity restrictions (54% of men and 62% of women aged 20 or over) (Table 2). But this question proportionally identifies more persons with functional limitations only than the being hampered or disabled question: 20% of these men and 16% of these women report only compensated chronic or functional problems and about 20% report uncompensated functional limitations without activity restrictions. We obtain roughly the same distribution for persons who regard themselves as having a disability.
25 From the aggregate functional health indicator, we can also estimate sensitivity and specificity scores for the being hampered or disabled question  Specificity measures the capacity of the being hampered...
suite. Table 3 reports these scores for three reference groups, based on different levels of the functional health indicator: (i) persons with disabilities in the broad sense, i.e., uncompensated functional limitations regardless of whether or not they affect activities (levels 3-5); (ii) only persons with activity restrictions (levels 4-5); (iii) only persons with severe activity restrictions (level 5). The being hampered or disabled question, asked at the household level, is very specific, but displays low sensitivity since it identifies few persons among those reporting disabilities of any sort: this group includes only 18% of persons having reported functional limitations or activity restrictions of all types. However, the sensitivity improves with the severity of the condition: the question identifies 27% of persons having reported activity restrictions of all types and about 50% of persons having reported restrictions in personal care activities. The long-term activity limitations or regard themselves as having a disability questions show higher sensitivity scores, although the concordance is again incomplete: for example, these questions identify 80% of persons who elsewhere report restrictions in personal care activities (Table 3).
Table 3 - Sensitivity (SE) and specificity (SP) of general questions on being hampered or disabled, long-term activity limitations and self-perceived disabilities measured against reported functional limitations and activity restrictions (%)
26 Age, gender, social status, and official recognition of a disability are characteristics that, for comparable functional problems, may affect the likelihood of identifying a disability situation via a general question (being hampered or disabled, feeling disabled, experiencing activity limitations). All other things being equal, and taking functional problems into account, the propensity to report being hampered or disabled is significantly greater among persons with little or no education, those who have obtained (or are awaiting) administrative recognition of a disability, the economically inactive (though not retired), and those who are or have been engaged in unskilled manual occupations (Table 4). The propensity of being identified as hampered or disabled is also significantly higher for men than for women when functional problems, age, socio-occupational category, and education are taken into account. However, this gender difference ceases to be significant with the introduction of the “administrative recognition of disability” variable: reporting administrative recognition, more common among men, is correlated with reporting being hampered or disabled, including when socio-demographic and health variables are taken into account.
Table 4 - Factors influencing the probability of reporting being hampered or disabled, long-term activity limitations and self-perceived disabilities by characteristic, for a given age and functional health status (age 30+)
27 In other words, besides the nature of functional problems, the characteristics that seem to induce respondents to report being hampered or disabled reflect specific situations that are conducive to revealing or “objectifying” a disability (economic inactivity, recognition of a disability, and so on). In the 2002-2003 survey, the reporting of being hampered or disabled may have focused on explicit functional problems, leaving less perceptible difficulties undocumented. This finding is consistent with the fact that the situations identified are, as a rule, fairly severe disabilities, which therefore tend to be recognized; and likewise with the fact that the question fails to identify many disability situations which are revealed by the aggregate functional health indicator. This result suggests that obvious impairments (such as visual impairment or amputation) or impairments qualifying for administrative recognition may be more routinely reported as being hampered or disabled, even when they apparently do not hinder activities (for example, because the person adjusts to his/her impairment). This may explain why a proportion of individuals identified by the being hampered or disabled question do not report activity limitations or do not regard themselves as having a disability.
28 We note that the propensity to report long-term activity limitations is less sensitive to socio-demographic variables than the propensity to report being hampered or disabled (Table 4): the two social-status variables (socio-occupational category and education, introduced separately or jointly) are not decisive. Only administrative recognition of a disability and – to a minor degree – age and being a woman significantly raise the propensity to report long-term activity limitations, when the other variables and functional health are taken into account. The question on self-perception of a disability (“do you regard yourself as having a disability?”) is sensitive to administrative recognition; some socio-occupational categories (farmers, homemakers) and educational attainment above middle school tend to raise the propensity to answer “regard myself as having a disability”, all other things being equal (the effect is borderline significant).
A loss of sensitivity for the being hampered or disabled question between the 1991-1992 and 2002-2003 surveys
29 Data from earlier Health Surveys do not enable us to repeat the analyses described in this study for the purpose of understanding the observed change. We can, however, use results from the question module on activity restrictions in the 1991-1992 survey, aimed at the sub-sample of people aged 65 or over (Robine et al., 1995; Cambois et al., 1996). Although the wordings and response categories do not allow a direct comparison of prevalences, the figures do not suggest a massive decrease in difficulties reported for personal care activities in the 1990s (four personal care activities are common to both surveys). Yet we see a very sharp decrease in the match between reported activity restrictions and reported being hampered or disabled. In the 1991-1992 survey, 82% of men and 74% of women aged 65 or over reporting restrictions in the four personal care activities were identified by the household question as being hampered or disabled (Robine et al., 1995); in the 2002-2003 survey, the percentages are 50% for men and 48% for women. The being hampered or disabled question in 2002-2003 thus identifies restrictions in personal care activities less frequently than in the 1991-1992 survey. This finding tends to corroborate the question’s selection effect and hence its lesser sensitivity, which may have accentuated the decline in prevalence.
30 Our study shows that the decrease in the prevalence of being hampered or disabled in the 1990s is far larger than that observed between 1980-1981 and 1991-1992, particularly at the oldest ages: the annual rate of change in prevalence for the over-65s is –1.6% between 1980-1981 and 1991-1992 and –5.3% between 1991-1992 and 2002-2003. In addition to a probable decrease in disability, we assume other factors have contributed to the change described by this indicator – first, because the changes at international level are not of comparable magnitude; second, because we observe a significant decrease in the number of people identified by this question who also report restrictions in personal care activities.
31 Longer life expectancy and improved healthcare systems may have brought changes in disability situations and self-perception of disabilities, thanks to better assistance for impairments and functional limitations for example. But we also assume that factors inherent to the Health Survey and changes in its protocol contributed to the decrease in persons who reported being hampered or disabled.
32 While better assistance in coping with difficulties and impairments may have actually reduced the number of disability situations, it may also have altered the perception of the term “disability” (handicap in French), restricting its current use to social disadvantages rather than to more common problems. As a result, the indicator may overstate the real decline in disability by making situations described as being hampered or disabled seem more radical than they actually are. Conversely, the overall improvement in health conditions may have raised expectations about health status, making functional problems that would have been deemed benign ten years previously appear more severe. If so, the propensity to report being hampered or disabled in equivalent situations should, if anything, increase. Likewise, better acceptance of a disability could lead to a more widespread reporting of it. These effects may both be at work, but they are difficult to tease apart as we have no objective measure of the change in functional health status. However, the loosening of the link between being hampered or disabled and reported restrictions in personal care activities (which are less sensitive to shifting perceptions) implies a less systematic reporting of certain disability situations via this general question. All in all, changing perceptions may have contributed to the decrease in prevalence.
33 Among the possible explanations for the sharp decline in reported being hampered or disabled, the first is the question’s change of position in the survey sequence. It now comes at the very beginning of the questionnaire, just after the socio-demographic module. The problem is that the prior listing of illnesses and medical consumption items in the questionnaire sequence – as in the 1991-1992 survey – probably increases the propensity to report disabilities by making respondents’ health problems more perceptible. In the 2002-2003 survey, the question “Do you regard yourself as having a disability?” is included in the third-visit questionnaire just after the module on functional limitations and activity restrictions. We find that the responses to the question are very strongly correlated with the reported functional limitations and activity restrictions enumerated previously, which can create an incentive effect. The prevalence is twice as high as for being hampered or disabled, despite the slightly less open-ended wording. This incentive effect was demonstrated in a study based on comparative results of performance tests and personal questionnaires from a single survey (Daltroy et al., 1999): for an equivalent functional health status, the prevalence of reported functional limitations is higher among persons who responded to the performance test before the questionnaire than among those who did so after. In the 2002-2003 survey, the being hampered or disabled question does not benefit from an enumeration effect or from the respondent’s growing familiarity with the survey and the interviewer, as may have been the case with the 1980-1981 and 1991-1992 surveys. This might explain why the reports are confined to proven disability situations and why the selection effect is stronger than in earlier surveys.
34 The household approach may also foster some reticence among respondents, particularly when they are asked to discuss disabilities (handicaps in French), a term which – as international studies emphasize – is negatively connotated and a cause of under-reporting. We lack the data needed to measure the impact of the interview method, but we can offer a few figures for discussion purposes.
35 The “Do you regard yourself as having a disability?” Question – asked on an individual basis in the 2002-2003 Health Survey – was included with the same wording in another survey on daily life and health (Vie Quotidienne et Santé, VQS), as a questionnaire appended to the 1999 population census  The “Daily Life And Health” questionnaire was used as...
suite. As in the Health Survey, the VQS question came after about ten questions on health and disabilities but, in the census context, it was designed to be answered by one or more persons on behalf of all household members. As the general contexts of the two surveys have little in common (and their timing was slightly different), we cannot effectively verify the existence and size of a collection-method effect. However, we can point to a far lower prevalence, with a strictly identical wording, in the VQS survey with its household approach: 9% of people aged 20 or over described themselves as having a disability in the 1999 VQS versus 15% in the 2002-2003 Health Survey (for people aged 65 or over, the respective proportions were 23% and 30%).
36 If the household collection method does have an effect on the reporting of disabilities, it could compound the other effects described earlier in the 2002-2003 survey, i.e. the change in perceptions and the question’s transfer to the very start of the survey. The household collection-method effect could also be strengthened by the Health Survey’s new general collection method. The respondent to this introductory part of the questionnaire is no longer responsible for the information on other members of his/her household, each of whom are interviewed separately for the remainder of the questionnaire. This may reduce the respondent’s incentive to mention difficulties or disabilities affecting other household members, except when they are manifest. Here as well, we lack the data needed to measure the respondent effect  The 2002-2003 questionnaire contains no variable for identifying...
37 In sum, none of these explanations fully accounts for the existence of a selection effect or its amplification in the 2002-2003 survey. We cannot effectively separate the effects of the wording that includes the term handicap from the question’s position in the survey, the questionnaire method, and the shift in perceptions; but their combination is likely to have impacted the way respondents answered the being hampered or disabled question in the 2002-2003 survey. This combination of effects may have contributed to the unexpectedly large decrease in the prevalence of being hampered or disabled and may help to explain the distinctive characteristics of the population group identified by this question in the 2002-2003 survey.
38 The being hampered or disabled question in the 2002-2003 Health Survey seems sensitive to a certain form of social recognition of disability. This recognition may be manifest (visible impairment), official (administrative recognition), objectified by the severity of the resulting difficulties (more often activity restrictions than functional limitations only) or linked to a particular social status (absence of qualifications, lack of skills, or economic inactivity). In this sense, the being hampered or disabled question in 2002-2003 reveals more cases of disadvantaged status than of functional problems. And this specific question shows, for example, that there are few gender differences here, despite the fact that women report a higher proportion of functional difficulties. The being hampered or disabled question is useful in its own right because of the phenomena that it reflects. But, in the light of the results of our study, we believe caution should be exercised in comparing prevalences of being hampered or disabled based on different surveys with the aim of studying their change over time, as the design of the latest survey questionnaire probably generates a selection effect.
39 This analysis underscores the importance of a survey question’s environment and wording. It also highlights the value of being able to compare responses to different questions in a given survey or from different data sources. First, such a comparison offers a fuller picture of the situation in a particular country and allows a more accurate interpretation of observed changes (Cambois et al., 2006). Second, it gives more detailed profiles of the populations identified, making it possible to select indicators appropriate to specific research objectives. Each question, with its particular wording, reflects a particular situation and set of needs. All the questions help to better understand the mechanisms that underlie the changing prevalence of functional limitations and their impact on activities and social participation.
40 * * *APPENDICES
Indicators, respondents and position of questions in surveys on health and medical care
Appendix 1 -
Wording of questions in 2002-2003 Health Survey used in our study
Appendix 2 -
Barberger-Gateau P., Rainville C., Letenneur L., Dartigues J.-F., 2000, “A hierarchical model of domains of disablement in the elderly: a longitudinal approach”, Disability and Rehabilitation, 22(7), pp. 308-317.
Bronnum-Hansen H., 2005, “Health expectancy in Denmark, 1987-2000”, European Journal of Public Health, 15(1), pp. 20-25.
Cambois E., Robine J.-M., Hayward M.D., 2001, “Social inequalities in disability-free life expectancy in the French male population, 1980-1991”, Demography, 38(4), pp. 513-524.
Cambois E., Robine J.-M., Romieu I., 1996, “Les types d’incapacités chez les personnes âgées”, Risques, (26), pp. 63-72.
Cambois E., Robine J.-M., Romieu I., 2005, “The influence of functional limitations and various demographic factors on self-reported activity restriction at older ages”, Disability and Rehabilitation, 27(15), pp. 871-883.
Cambois E., Clavel A., Robine J.-M., 2006, “L’espérance de vie sans incapacité continue d’augmenter”, Solidarité Santé, no. 2, pp. 7-22.
Crimmins E., 2004, “Trends in the health of the elderly”, Annual Reviews of Public Health, 25, pp. 79-98.
Daltroy L.H., Larson M.G., Eaton H. M., Phillips C.B., Liang M.H., 1999, “Discrepancies between self-reported and observed physical function in the elderly: the influence of response shift and other factors, Social Science & Medicine, 48(11), pp. 1549-1561.
Fougeyrollas P., 1995, Le processus de production culturelle du handicap. Contextes sociohistoriques du développement des connaissances dans le champ des différences corporelles et fonctionnelles, Quebec, CQCIDIH-SCCIDIH.
Freedman V.A., Crimmins E., Schoeni R.F., Spillman B.C., Aykan H., Kramarow E., Land K., Lubitz J., Manton K., Martin L. G., Shinberg D., Waidman T., 2004, “Resolving inconsistencies in trends in old-age disability: report from a technical working group”, Demography, 41(3), pp. 417-441.
Freedman V.A., Martin L.G., Schoeni R.F., 2002, “Recent trends in disability and functioning among older adults in the United States: a systematic review”, JAMA, 288(24), pp. 3137-3146.
Guralnik J.M., Ferrucci L., 2003, “Assessing the building blocks of function: utilizing measure of functional limitations”, American Journal of Preventive Medicine, 25(3 sup. 2), pp. 112-121.
Jagger C., Arthur A.J., Spiers N.A., Clarke M., 2001, “Patterns of onset of disability in activities of daily living with age”, Journal of American Geriatric Society, 49(4), pp. 404-409.
Lagergren M., Batljan I., 2000, “Will there be a helping hand: macroeconomic scenarios of future needs and costs of health and social care for the elderly in Sweden, 2000-2030”, Annex 8 to the Long-term survey 1999/2000, Stockholm, Ministry of Health and Social Affairs.
Mormiche P., Groupe de projet HID, 2000, “Le handicap se conjugue au pluriel”, Insee Première, no. 742, 4 pp.
Ravaud J.-f., Delcey M., Yelnik A., 1999, “Construct validity of the functional independence measure (FIM): questioning the unidimensionality of the scale and the ‘value’ of FIM scores”, Scandinavian Journal of Rehabilitation Medicine, 31(1), pp. 31-41.
Ravaud J.-f., Letourmy A., Ville I., 2002, “Identifying the population with disability: The approach of an INSEE survey on daily life and health”, Population-E, 57(3), pp. 529-552.
Robine J.-m., Cambois E., Romieu I., 1995, Espérances de santé pour la population âgée de 65 ans et plus: application des concepts de la classification internationale des handicaps aux calculs d’espérances de santé, Final report to MIRE, Montpellier, November 1995, 95 pp.
Robine J.-m., Mormiche P., 1994, “Estimation de la valeur de l’espérance de vie sans incapacité en France en 1991”, Solidarité Santé, no. 1, pp. 17-36.
Robine J.-M., Mormiche P., Sermet C., 1998, “Examination of the causes and mechanisms of the increase in disability-free life expectancy”, Journal of Aging and Health, 10(2), pp. 171-191.
Robine J.-M., Ravaud J.-F., Cambois E., 1997, “General concepts of disablement” in Hamerman D. (ed.), Osteoarthritis: Public Health Implication for an Aging Population, John Hopkins University Press, pp. 63-83.
Robine J.-M., Romieu I., Cambois E., 1999, “Les indicateurs d’espérance de santé”, Bulletin of the World Health Organization, Recueil d’articles 1, pp. 107-111.
Robine J.-M., Romieu I., Michel J.-P., 2003, “Trends in health expectancies”, in Robine J.-M., Jagger C., Mathers C., Crimmins E. and Suzman R. (eds), Determining Health Expectancies, Chichester, John Wiley & Sons Ltd, pp. 75-101.
Schoeni R., Freedman V., Martin L., 2005, “Socioeconomic and demographic disparities in trends in old-ages disability”, Trends Working Paper 05-1 (http://trends.psc.isr.umich.edu).
Simonsick E.M., Kasper J.D., Guralnik J.-M., Bandeen-Roche K., Ferrucci L., Hirsch R., Leveille S., Rantanen T., Fried L.P., 2001, “Severity of upper and lower extremity functional limitation: scale development and validation with self-report and performance-based measures of physical function. WHAS Research Group. Women’s Health and Aging Study”, Journal of Gerontology: Psychological Sciences, 56B(1), pp. S10-19.
Verbrugge L.M., Jette A.M., 1994, “The disablement process”, Social Science & Medicine, 38, pp. 1-14.
Who (World Health Organization), 1980, International Classification of Impairments, Disabilities, and Handicaps, Geneva (reprinted 1993).
Wood P.H.N., Badley E. M., 1978, “An epidemiological appraisal of disablement” in BENNETT A. E. (ed.), Recent Advances in Community Medicine, Edinburgh, Churchill Livingstone, pp. 149-173.
[ *] Institut national d’études démographiques, Paris.
[ **] Institut national de la santé et de la recherche médicale, Montpellier.
[ ***] Institut national de la statistique et des études économiques, Paris. Translated by Jonathan Mandelbaum.
[ 1] The modules on functional limitations and activity restrictions were administered only to adults, which is why we confine our analysis to these ages. However, for the being hampered or disabled question, which concerned all household members, we report the prevalence for all ages, including for children.
[ 2] Specificity measures the capacity of the being hampered or disabled question to identify persons actually affected by functional problems (so as to limit the number of “false positives”); sensitivity measures the question’s capacity to identify all persons with functional problems (so as to limit the number of “false negatives”). While the aggregate functional health indicator is not, strictly speaking, an objective measure of disabilities, it does enable us to define groups of persons experiencing different types of disability situations, and we use these groups as reference populations to estimate the sensitivity and specificity of the general question.
[ 3] The “Daily Life And Health” questionnaire was used as a filter for selecting the HID survey sample. It was added to 400,000 forms in the 1999 population census and included some twenty questions on disability.
[ 4] The 2002-2003 questionnaire contains no variable for identifying the respondent to the being hampered or disabled question. Such a variable would have made it possible to determine whether the decrease in prevalence is smaller in the sample of persons responding on their own behalf. Moreover, the analyses on this topic using 1991-1992 data are inconclusive because of the small sample size (Robine et al., 1995).
L’enquête Santé française posait en 2002-2003, pour la troisième fois depuis 1980, une question générale sur les « gênes ou handicaps dans la vie quotidienne ». Les réponses obtenues suggèrent une décroissance beaucoup plus rapide qu’auparavant des prévalences de l’incapacité au cours de la dernière décennie, dont l’ampleur dépasse les scénarii les plus optimistes. Ce travail examine plus particulièrement les changements intervenus dans le protocole de la dernière enquête et leur influence possible sur la comparabilité des résultats avec ceux des enquêtes antérieures. L’analyse montre que la question générale de l’enquête de 2002-2003 enregistre surtout des restrictions d’activité sévères et des situations de handicap avérées, donc relativement peu répandues, alors que la formulation visait à recueillir aussi des situations de « simples gênes », bien plus fréquentes. Cette sélection a été plus prononcée dans la dernière enquête que dans les enquêtes précédentes. Au total, la question sur les gênes ou handicaps ne permet pas de poursuivre la série chronologique initiée en 1980, mais elle vient compléter la connaissance des différents types de problèmes fonctionnels à travers les situations de désavantage social qu’elle semble plus spécifiquement refléter.
In 2002-2003, the French Health Survey included – for the third time since 1980 – a general question on “being hampered or disabled in daily life”. The responses obtained suggest that the prevalence of disability has fallen far more rapidly in the past decade than in earlier periods. Indeed, the drop exceeds the most optimistic scenarios. Our study devotes special attention to changes in the survey protocol and their possible influence on comparability of the latest results with those of previous surveys. The analysis shows that the general question in the 2002-2003 survey mainly records severe activity restrictions and proven (and thus relatively rare) disability situations, whereas the wording was designed to identify “basic limitations” as well, which are far more common. This selection was more pronounced in the latest survey than in its predecessors. In conclusion, the being hampered or disabled question does not provide continuity with the time series begun in 1980, but it yields additional information on the various types of functional problems via the situations of social disadvantage that it more specifically brings to light.
La encuesta sobre la Sanidad francesa planteaba en el bienio 2002-2003, por tercera vez desde 1980, una pregunta general sobre las « molestias o dificultades en la vida cotidiana ». Las respuestas obtenidas sugieren una disminución mucho más rápida que anteriormente de los casos de incapacidad a lo largo del último decenio, cuyo alcance supera las previsiones más optimistas. Este análisis examina más concretamente los cambios ocurridos en el protocolo de la última encuesta y su posible influencia sobre la posibilidad de comparar los resultados con los de las encuestas anteriores. El análisis muestra que la pregunta general de la encuesta de 2002-2003 revela sobre todo graves restricciones de actividad y situaciones de discapacidad comprobadas, y por consiguiente relativamente poco extendidas, cuando en realidad la formulación buscaba también situaciones de « simples molestias », mucho más frecuentes. Esta selección ha sido más pronunciada en la última encuesta que en las encuestas anteriores. En resumen, la pregunta sobre las molestias o discapacidades no permite prolongar con la serie cronológica iniciada en 1980, pero permite completar el conocimiento de los distintos tipos de problemas funcionales a través de situaciones de desventaja social que las respuestas a esta pregunta paracen reflejar más específicamente.
PLAN DE L'ARTICLE
- I - Context of our study
- II - Disabilities in the Health Surveys
- III - What does the being hampered or disabled question measure in 2002-2003?
- IV Discussion
POUR CITER CET ARTICLE
Emmanuelle Cambois et al. « Did the Prevalence of Disability in France Really Fall Sharply in the 1990s? », Population (English Edition) 2/2007 (Vol. 62), p. 313-337.
URL : www.cairn.info/revue-population-english-2007-2-page-313.htm.
DOI : 10.3917/pope.702.0313.