This article presents the objectives and methodology of a French study on aging, health, and work called VISAT (Vieillissement, Santé, Travail), an ongoing prospective research project. It was designed to determine how and to what extent work conditions (types of tasks performed, work organization, exposure to occupational hazards, and other work-related conditions) favorably or unfavorably affect the availability of the individual’s physiological and psychological resources, in relation to the effects of aging on those resources. Very few studies on the health-work relationship have focused on age, and even fewer have done so in a longitudinal way. One exception is the health, work, and aging study called ESTEV (Étude Santé, Travail et Vieillissement; Derriennic, Touranchet, & Volkoff, 1992). Although its approach is basically the same, the VISAT study nevertheless deviates from ESTEV in some ways. The participant pool for the VISAT study is smaller (3237 as compared to 22,000 participants in the original ESTEV study), but its approach is more in-depth (more basic data on each worker). It has at least two other specific features as well, including (i) a deeper grasp of psychological components (perceptual, cognitive, affective, and psy- chosocial) related to both job demands and the resources of the individual, and (ii) investigation of the effects of work conditions after retirement as early as the first cross-sectional step.

Due to its complexity, the relationship between aging, health, and work cannot be addressed as a whole. For this reason, the VISAT database was designed to be acessed by several research teams from various disciplines (e.g., epidemiology, occupational medicine, occupational health psycho- logy, psychology of aging, psychosociology), and to permit examination of several specific components of the relationship. This was the rationale for collecting a broad set of data in VISAT. In this paper, we first present the study’s main strategic choices, which resulted both from research goals and practical constraints, and are common to all planned investigations. We then give some examples of elementary research projects and their theoretical background in order to illustrate how the study intends to shed light both on fundamental research issues and on prevention policies. Finally, the various aspects of the methodology are described.

One goal of VISAT is to shed light on several aspects of this issue. For instance, beyond differences between jobs, are there certain characteristic of work activities that are more specifically associated with certain sleep disorders (e.g., heavy physical loads, time pressure, tension resulting from social interactions with customers or partners)? What kinds of relationships exist between self-assessed sleep quality and various dimensions of perceived health, as it is measured by the Nottingham Health Profile, for instance (Bucquet, Condon, & Ritchie, 1990)? Earlier observations suggest that reported sleep difficulties may be a sign that the individual is no longer able to cope with stress in his/her life. Do people who have worked on shifts and are now back to normal work schedules exhibit different sleep difficulty patterns from current shiftworkers and from persons who have never worked on shifts? In other words, are there permanent effects of shiftwork on sleep? Some earlier findings support this view (Barrit, Brugère, Butat, Cosset, & Volkoff, 1994), while others do not (Marquié & Foret, 1999; Webb, 1983). The VISAT study can contribute to explaining such inconsistencies, not only by looking at factors that have either rarely or never been taken into account in a same study (duration of shiftwork experience, time since the individual stopped working on shifts, the working-life period concerned), but also by poviding longitudinal data. Physiological differences between males and females do not seem to be able to account for the significantly higher frequency of the reported sleep disorders always observed. The fine-grained analysis of working conditions made possible by VISAT is a means of clarifying this issue. Finally, the presence of retirees in our sample allows us to examine which sleep troubles are age-dependent (that is, expected not to be influenced by retirement) and those which are work-dependent (that is, expected to be less frequently reported by retirees than by people still working).

Another subgoal of VISAT concerns the joint effects of aging and working conditions on cognitive resources. As with other personal resources, the links between work experience and cognitive capabilities in relation to age are complex and have been widely neglected until now both in the gerontological research and in occupational psychology. Yet there is evidence that various non-cognitive and cognitive work experiences, which are addressed in our study, influence the intellectual functioning of individuals. They may even have long-term effects on cognitive resources, and thereby modulate the effect of aging on those resources. For instance, poor cognitive activities in certain jobs have often been suspected to generate some forms of cognitive sclerosis (e.g., Leplat, 1988; Schooler, 1984). Based on longitudinal data and the use of appropriate statistical models, Schooler, Mulatu, and Oates (1999) found a reciprocal but independent relationship between intellectual functioning and the cognitive complexity of the work environment. Their findings suggest that substantively complex work significantly increases intellectual flexibility, as well as other dimensions of cognition, and that this effect is even greater among older workers. A similar view can be found in the so-called differential preservation or maintenance hypothesis, which is implicit in many studies on the psychology of aging. According to this hypothesis, intensive and continuous use of basic cognitive abilities during adulthood would prevent those abilities from undergoing the usual age-related decline (for a review, see Marquié, 1997). Recent work in neuroscience supports this hypothesis and shows that adult brain plasticity turns out to be greater than expected. One example is found in the study on taxi drivers by Maguire et al. (2000). This study revealed structural changes in the hippocampus, a brain region known to be involved in navigation, that were correlated with the amount of time spent as a taxi driver. Data collected in the VISAT research program, especially information about job-related cognitive demands and the efficiency of several cognitive functions, should provide further evidence for or against the maintenance hypothesis. Work experience may also affect intellectual functioning via the more or less positive self-representation of one’s abilities that they contribute to generating, which itself is known to be linked to motivational factors. For instance, the effort invested in a task has been shown to be related to self-efficacy beliefs (Pajares, 1997). In order to examine this hypothesis, data about memory self-efficacy beliefs were recorded in the VISAT project. More positive self-efficacy beliefs are expected to be associated with cognitively complex and demanding jobs, and with higher memory performance, after adjustment for possible confounding variables such as education, age, and sex.

Health and biological factors are another possible mediator of underlying the relationship between occupational experience and intellec- tual functioning at different ages. Specific factors (such as exposure to organic solvents) are suspected to have long-term effects on cognitive efficiency (Baker, 1994; Lees-Haley & Williams, 1997). Much work still needs to be done to further document these effects. However, beyond the separate effects of this type of factor, it is the cumulative effect of several minor as well as more serious work-related health impairments, such as sensory losses, nervous or cardiovascular problems, sleep disorders, and other biological life events, that should be evaluated, because they may account for a significant part of the total observed age-related variance in the availability of cognitive resources (Elias, Elias, & Elias, 1990; Houx, Vreeling, & Jolles, 1991). This is part of the VISAT program.

Other research projects based on current theoretical and social issues on aging and health at work also determined the VISAT design and materials. This is true of the study of the still poorly understood association between psychosocial hazards at work and adverse health outcomes, as it is accounted for in the job demand control (Karasek & Theorell, 1990) and effort-reward imbalance (Siegrist, 1996) models. Expected contributions concern the issue of the independency between the two models, the relationships between the model components and occupational status, and the role played by both the duration and the time (age) of exposure in the appearance of health problems during adult life. A final example concerns the effects of economic insecurity (loss of job or income) (Catalano, 1991). The data collected, especially follow-up data, should allow us to specify its effects on several aspects of health, especially in relation to the worker’s age (Gallo, Bradley, Siegel, & Kasl, 2000).

A training course was offered to all participating physicians. It was designed to enable them, as interviewers, to achieve maximal standar- dization of the inquiry conditions. This was especially important for the parts of the protocol that are less familiar to physicians, such as psychometric testing. A guidebook explaining the interviewing techniques was provided along with the other necessary survey documents.

As stated above, the goal of VISAT was not to obtain a representative sample of the total adult population. However, the overall distribution by sex and socioprofessional category in our sample (presented in Table 1) was very close to that observed at the national level (INSEE, 1996). For the employees born in 1934 who were still working, the distribution across the two sexes differed notably from the retired persons in the same age group, with men being underrepresented. Moreover, in this same group, executives, and persons with high-ranking intellectual occupations, as well as office staff, were overrepresented in comparison to technicians/supervisors and blue-collar workers.

TABLE 1 :
Distribution (%) of participants, by year of birth socioprofessional category, sex, and current work status (still working/retired)
Distribution (%) des participants en fonction de l’année de naissance, de la catégorie socioprofessionnelle, du sexe, et de l’activité (actifs/non actifs)

The participation rate in 1996 was 76 %. A survey conducted a posteriori on a random sample of the occupational physicians revealed a variety of causes for non-participation in the study (e.g., refusal, overload of work, work hours incompatible with testing schedule, business trips, recently left the company, maternity leave, and sick leave), and no notable difference between the persons who participated in the study and the ones who did not. In the older group, 83 % had left the work force definitively, and most of these individuals had stopped working less than five years earlier (84.3 %). The main reasons for no longer working were normal retirement (age 60 or slightly younger, 79.9 %), resignation or dismissal (10 %), and an accident or disease (6.9 %).

The next question consisted of 24 items aimed at describing the person’s feeling of self-accomplishment and room to maneuver at the current job. One series of items assessed the person’s sense of achievement relative to the level attained by his/her parents, the choice of one’s occupation and motivation to keep the same job until retirement, the level of perceived qualification, the opportunity to pass one’s experience on to younger co-workers, and potential recognition obtained from other people at work. Another series covered the sense of control over one’s work situation afforded by the means made available to the employee, including autonomy, responsibilities granted, adequate equipment and supplies for performing one’s job, and support from others in the work environment. The remaining items asked the participants to assess their chances of mobility (hireability by other companies or for other jobs) and their potential for career advancement and skill enhancement. Many items explicitly referred to the decision latitude dimension of Karasek and Theorell’s (1990) model and the effort-reward imbalance dimension in Siegrist’s (1996) model. However, only proxy measures of these models could be used because validated French versions of the original instruments did not exist at the time the study was designed. For each item, the participants had to say whether the concerned work characteristic was applicable to them by respon- ding on a four-point Likert scale. Following this 24-item series, the respondents were asked to assess their overall satisfaction with their job on a 10-point scale ranging from “Not satisfied at all” to “Fully satisfied”.

Another question, composed of 25 items, was designed to measure any difficulties the person was encountering in his/her current position. The difficulties in question were related to physical exertion, physical environment at the work place, whether the work schedule was socially and biologically out of phase, demands and requirements of various types including perceptual (e.g., “Detecting fine details”), motor (e.g., “Making very precise movements”), attentional (e.g., “Being required to constantly keep one’s eyes on one’s work”), memory-related (e.g., “Having to simultaneously retain large amounts of information”), affective and psychosocial (e.g., “Having to deal with the public” or “Having to cope with the suffering of others”), occupational risk-related, and fear of losing one’s job. The participants had to indicate whether or not the difficulty “concerned” them; if so, they further indicated whether or not they experienced it as “Very difficult” and whether or not it was “Harder to handle with age”.

In another question, the participants had to state the principal characteristics of the work day and the quality of the subsequent sleep period right before the testing. The next question was made up of items from the French version of the Nottingham Health Profile (NHP) validated by Bucquet et al. (1990). It had 17 items corresponding to three of the six dimensions of the NHP: energy level and stamina (three items), emotional reactions (nine items), and social isolation (five items). After that, the participants made an overall evaluation of their state of health on a ten-point scale ranging from “Very bad” to “Very good”. Then they were asked to indicate their perceived age relative to their real age, again on a ten-point scale ranging from “Much younger than my age” to “Much older than my age”. Finally, the retirees had to use a four-point scale to assess the effects of retiring on how they felt in general, their sense of social utility, and their attitude about their future.

The clinical measures from the medical examination consisted of blood pressure, heart rate, height and weight, spinal column flexibility, and muscle strength measured on a dynamometer. Supplementary tests were performed when the necessary equipment was available. They included a respiratory examination (respiratory capacity and maximum air volume exhaled per second), visual acuity for near and far vision (in each eye), and hearing loss in each ear for six frequencies ranging from 500 Hz to 6000 Hz. In addition to their merits in terms of feasibility, the utility of these physiological measures lies in their potential ability to act as markers of aging, as well as their sensitivity to working conditions (Derriennic et al., 1989).

More specifically, the word-list learning test consisted of three trials, each followed by immediate recall. This test was an adapted version of the Rey auditory verbal learning test, but with only three consecutive learning trials instead of five, and a list of 16 words instead of 15. Performance on this type of test turned out to be a sensitive measure of age-related learning differences (e.g., Geffen, Moar, O’Hanlon, Clark, & Geffen, 1990; Mitrushina, Satz, Chervinsky, & D’elia, 1991). Given the longitudinal nature of the study, three different 16-word lists (A, B, and C) were set up so the participants would be tested five and then ten years later on different words each time. Another difference from the Rey test was the choice of words. On our lists, high-imagery words alternated with low-imagery words. The high- and low-imagery words were selected on the basis of a study of the imagery value of more than 1000 frequent spoken French nouns (Hogenraad & Orianne, 1981). The words were two-syllable, phonetically unambiguous common nouns.

The next test was the WAIS (1970) digit symbol substitution subtest, which is considered to be highly loaded by the information processing speed component and very sensitive to aging effects (Salthouse, 1992). The selective attention test that followed was derived from the Sternberg’s test (1975) and was composed of two subtests. The first was a task consisting in looking as quickly as possible through a line of 58 alphabetic characters (57 of which were distractors) to find a target letter shown in the margin, and then crossing it out. This task was repeated six times, on six lines with a different target and different distractors each time. The location of the target letter among the distractors was random. The second subtest also had six lines of 58 alphabetic characters, but this time the memory load was greater because the target to locate was one of the four letters shown in the margin. In the next test, delayed recall, the participants had to write down all the words they could remember on the sheet of paper provided. Finally, the recognition task involved finding the 16 learned target words randomly mixed in with 32 fillers. The same rules were used to generate the filler words for this task as for the test words. As in the learning task, three versions of the recognition test materials were generated.

All tests were run by the physician, who had been given the necessary training. Other procedures for the cognitive tests were as follows. For the learning and recognition tests, one of the three lists, A, B, or C, was randomly assigned to each new participant. In the learning test with immediate recall, the words on the list were first read aloud by the experimenter in a clear and carefully articulated voice, at a pace of one word per second. Immediately afterwards, the participant was asked to recall the words in any order, but the recall order as well as any intrusions were noted. Then the second and third trials took place under exactly the same conditions (reading by experimenter followed by immediate recall). There were no time limits on either type of recall (immediate or delayed) or on the recognition task. On the other hand, in the digit symbol substitution test and both parts of the selective attention test, the instructions stressed both speed and accuracy. Each selective attention subtest was limited to five minutes. For the digit symbol substitution test and both selective attention subtests, a demonstration and a practice trial were given beforehand. For the delayed recall test also, the words could be recalled in any order.

Data collection was standard and usually took place in the same order used above to present the materials. The occupational questionnaire was filled out by the employee him/herself in most cases, although help was provided if necessary by a medical assistant who was careful to guide the participants without influencing their answers. Next, the employee was received by the occupational physician, who filled out the medical questionnaire, and carried out the cognitive tests. Then the third questionnaire about the participant’s personal life was filled out by the participant outside the physician’s office, again, with the help of the assistant if needed. Finally, the supplementary tests (respiratory, visual, and auditory) were performed by the assistant. The total testing time for the employee averaged an hour and a half.

Paper received: june 2001.

Accepted by G. Karnas in revised form: february 2002.