Participating in life situations "as and when you want" is an essential part of life, and establishing the natural history of person-perceived participation restriction in older adults provides a perspective on the effects of health on everyday life and the potential need for effective interventions to improve participation. Estimating incidence and recovery highlights the potential for participation restriction to be modified or prevented. We have previously reported that about one half of a general population sample of adults aged 50 and over report participation restriction at any one time; now in this prospective follow-up of these community-dwelling older adults, we have shown that there is a substantial degree of change in participation status over a three-year period. Nearly 30% of those who were participating "as and when they wanted" in all aspects of life at baseline indicated that they were not doing so in at least one aspect of life three years later, whereas almost one third of those who had a restriction at baseline were reporting at three-year follow-up that they were now free of restrictions in any aspect of their lives. In addition many of those who continue to indicate restriction in at least one aspect of life at three years, have indicated recovery and onset of restriction in different areas to those indicated at baseline. However these figures must be set in the context that, for most people, overall status remained unchanged; 39% of the sample remained restriction free at three year follow-up and most persons reporting a restriction at baseline were still restricted in at least one aspect of life three years later. Although the three-year interval may miss meaningful transition between participation and participation restriction [12, 15], our findings still highlight that participation status over this time period changes for some adults and remains stable for others.
These overall figures conceal substantial contrasts in the patterns of change with age. The high likelihood of persistence at older ages combined with the age-related increase in the 3-year onset underlines the cumulative problem of participation restriction as people become older – exemplified by the declining likelihood of recovery as people reach the oldest ages. Both incidence and persistence are also higher in women, especially in women aged 80 years and over, where half of those without restriction at baseline indicated at least one aspect of life restricted by three year follow-up, and 90% of women in this age-group with a baseline restriction continued to indicate restriction at three-year follow-up.
Older adults who were already experiencing restriction in one aspect of life at baseline were more likely to indicate restriction in other areas at three year follow-up than those free of restriction. This indicates that once restriction is present in one area of life, this becomes a risk factor or risk marker for further areas of life to become restricted. By contrast there is evidence that restriction is reversible for some people, notably when it occurs in men and adults in the younger age range of our sample of over-50 year olds, and when it is an isolated problem in one aspect of life. However these rates of recovery may be over-estimated for each aspect of life because they will include individuals who no longer wish to participate in that aspect of life, either because it has become too difficult to maintain participation (in which case the extent of persistence will be under-estimated) or because they simply no longer need to participate in such areas. This may be particularly true for work, education and looking after dependents, where persistence was lowest and the proportion with participation restriction decreased with age .
The majority of the onset of restriction in mobility outside the home, managing money, work, education or social activities is experienced in those with no restriction in any other aspect of life and this indicates how the profile of participation restriction begins (data not shown). These aspects of life provide potential priority areas in which early interventions may prevent further restriction. In this model, the evidence presented here highlights again the prime importance of mobility outside the home because it is the most common form of restriction and problems with mobility are often the first function to become limited in the disability process .
The estimates of frequency of onset and persistence of participation restriction, adjusting simultaneously for age, gender, socio-economic status and physical and mental health, in persons who were subsequently lost to follow-up were similar to those observed for persons who completed the KAP at three years. By contrast clear differences in age, gender, socio-economic status and general health were observed between those included in the analysis and those who had died at follow-up. Although prevalence declined in the three-year period because of this selective attrition, the overall number of people with participation restriction in the sample followed from baseline to three years has increased because the number of people who have indicated onset of participation restriction is greater than the number who indicated recovery.
There is a possibility that recovery may also be over-estimated due to regression to the mean which occurs when apparent abnormalities on initial investigation have a high probability of being at the extreme end of an individual's normal range because of random variation  or responder bias (people who recover may be more likely to respond to a follow-up questionnaire). However participation restriction at baseline was higher in persons who subsequently did not respond and it seems unlikely that recovery estimates are substantially biased. Alternatively some of the changes may be due to measurement error rather than a true change in participation, although repeatability investigations had suggested that most items on the KAP are not subject to large responder variations in short-term reporting .
Patterns of incidence and persistence over time for individual aspects of life varied with age and gender and were not necessarily the same as those cross-sectional relationships observed at baseline between prevalence, age and gender . For example, at baseline the prevalence of restricted self-care increased with age, whereas the persistence of restricted self-care was higher in the younger age-groups (50–59, 60–69, 70–79) than in those aged 80 and over. In addition to the limitations of this study discussed previously (attrition, measurement error and regression to the mean), small numbers of those restricted at baseline for some individual items (such as self-care) and the consequent unstable estimate of change, may explain these inconsistent relationships between age and restriction patterns over time in some aspects of life.
We have previously shown that person-perceived participation restriction is common in adults aged 50 years and over in the community but the follow-up study has highlighted the substantial degree of change in participation status that can occur in a three-year time period. The results indicate the potential for prevention and reduction in the level of restriction. Prevention of onset or early intervention to reverse isolated restriction could potentially reduce the risk of progression to other areas of restriction. Even for persons who have established restrictions, the observations of a significant rate of reversibility underlines the need to identify ways to enhance such improvement.
Our results reflect previous studies of the changes and extent of limitations in other forms of disability [11, 12]. Deeg  reported that 53% of adults aged between 55 and 85 did not report limitations in climbing stairs, cutting toenails or using transportation over a six-year period. Barberger-Gateau and colleagues  reported that limitation in three definitions of disability (activities of daily life, instrumental activities of daily life, mobility and housework) persisted in around 50% of adults aged 65 and over who had the particular definition at baseline. For participation, the causes of the onset, persistence and recovery are likely to be numerous. Although medical conditions play a major role, and medical interventions may directly improve participation, environmental influences on functioning may be of greater importance in older adults, such that participation may be maintained even in the presence of health conditions, impairments and activity limitations . In future analyses we will investigate the role of such influences on change in participation status.