Article Text

Download PDFPDF
Projecting long-term trends in mobility limitations: impact of excess weight, smoking and physical inactivity
  1. Tommi Härkänen1,
  2. Päivi Sainio1,
  3. Sari Stenholm2,
  4. Annamari Lundqvist1,
  5. Heli Valkeinen1,
  6. Arpo Aromaa1,
  7. Seppo Koskinen1
  1. 1 National Institute for Health and Welfare, Helsinki, Finland
  2. 2 Department of Public Health, University of Turku, Turku, Finland
  1. Correspondence to Dr Tommi Härkänen, National Institute for Health and Welfare, PO Box 30, Helsinki FI-00271, Finland; tommi.harkanen{at}thl.fi

Abstract

Background Policy makers need disability projections for planning adequate services and measures for health promotion. The aim of this study is to provide projections on severe mobility limitations up to year 2044 and illustrate how the projected prevalence and the number of persons with severe mobility limitations are affected by potential changes in the modifiable risk factors, namely excess weight, physical inactivity and smoking.

Methods We analysed the nationally representative, repeated measures Health 2000 and 2011 Surveys (BRIF8901) with 8615 and 6740 participants, respectively, aged 18 years and older. Severe mobility limitations were defined as major difficulties or unable to walk about half a kilometre. We applied a multistate model on repeated measures to account for both individual risk factors and their changes over time.

Results The number of people with severe mobility limitations was projected to double by the year 2044 in Finland, due to the rapid ageing of the population. Eliminating half of the excess weight would reduce their number by one-fifth, while reductions in the prevalence of smoking and physical inactivity would have a minor impact. Even if excess weight, smoking and physical inactivity were completely eliminated, the number of persons with severe mobility limitations is projected to increase.

Conclusions Designing and implementing strategies to promote healthy weight are important to slow down the rapid increase in mobility limitations due to population ageing. Providing adequate health and social services for the increasing population with disabilities will nevertheless be an increasing national challenge.

  • ageing
  • longitudinal studies
  • obesity
  • public health
  • physical function

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Understanding population-level changes in functioning is of crucial significance because the population is rapidly ageing throughout the world.1 This information is useful in making policy decisions, for example, incentivising health behaviours or increasing rehabilitation services. A central part of functioning is mobility, which is a prerequisite to participation in civic life as well as an important component of quality of life.2 Restricted mobility also predicts limitations in daily activities, morbidity and falls, institutionalisation and mortality.3 4

Studies on long-term time trends demonstrate a decrease in mobility limitations in the 1980s and the early 1990s that levelled off and then reversed.5 6 Recently, increases in mobility limitations among older adults have been reported in many countries7–11 and also in the working-aged population in the USA.7 12 13 The causes behind this increase are still debatable, but increasing obesity has been suggested to have a notable influence.14 15 The reversal of the favourable disability trends has raised a deep concern about future financing and provision of adequate services for older people (eg, ref 10).

The future development of disability will largely depend on changes in causative factors. Changes in the modifiable risk factors of mobility limitations explain a large part of the past changes in mobility.7 14 16 17 Thus, their development is of crucial importance also in projecting future changes in mobility. The most essential modifiable risk factors affecting mobility limitations include excess weight, smoking and physical inactivity.18–20

Several projections indicate an increase in the disability prevalence and the number of disabled persons over the next decades.21–25 However, the outcome has often been a quite heterogeneous set of functional limitations, and relevant modifiable risk factors have not always been included. Moreover, individual changes in risk factors over time have generally not been taken into account, nor different risk factor distributions in birth cohorts and population subgroups. Furthermore, earlier projections have neglected to account for the impact of risk factors on mortality and to consider different sources of uncertainty.

By applying multistate modelling26 to individual-level follow-up survey data, this study aims to overcome the limitations in earlier projections and to: (1) provide projections on mobility limitations up to the year 2044 and (2) illustrate how the projected prevalence of mobility limitations as well as the population size and age structure are affected by potential changes in three modifiable risk factors, namely excess weight, physical inactivity and smoking. We also incorporate the different sources of uncertainty inherent in all projections.27 28

Methods

Data

The Health 2000 Survey (BRIF8901) was a nationally representative comprehensive health examination survey with assessments, interviews and questionnaires conducted in Finland in 2000–2001. The two-stage stratified cluster sample consisted of 9922 persons aged 18 years or older. The participation rate was 92%.29

Eleven years later, all members of the Health 2000 sample, now aged 29 years or over, who had not refused further contacts, and lived in Finland, were invited to participate in the follow-up, the Health 2011 Survey resulting in a repeated measures design.30 31 A new sample (n=1994) of younger adults (aged 18–28 years) was also included. Altogether 10 129 persons were invited and 67% participated. The data collection is described in the online supplement.

Supplementary file 1

The participants signed a written informed consent approving also record linkages.

Variables measured in 2000 and in 2011

Outcomes

Walking was chosen as the indicator of mobility, because it enables people living in the community to accomplish many everyday activities, such as grocery shopping, reaching the bus stop or visiting neighbours,4 and therefore, problems in walking contribute to disability burden. Mobility was assessed with the question ‘Are you able to walk about half a kilometer without resting?’ with response options, ‘with no difficulties, with minor difficulties, with major difficulties, and not at all’.32 The two latter were combined to indicate severe mobility limitations.

Mortality follow-up on all-cause mortality obtained from Statistics Finland continued until 30 June 2012 and was individually linked with the survey data.

Predictors

Body mass index (BMI) was based on measured height and weight. Self-reports were used if measures were missing. BMI was used as a continuous measure in the modelling. In table 1, we report the transition probabilities using the dichotomy non-obese versus obese (BMI ≥30 kg/m2). Physical inactivity was based on the question, ‘How much do you exercise and strain yourself physically in your leisure time?’ with four response options describing the strenuousness of exercise. Smoking was based on the question, ‘Do you smoke nowadays?’. The response options ‘daily’, ‘occasionally’ and ‘no’ were dichotomised as daily versus no or occasionally. Level of education was classified as low, intermediate and high. More details of the variables are presented in the online supplementary table S9.

Table 1

Crude transition probabilities (%) from one risk factor category to the other or to death, during the follow-up from year 2000 to 2011 and number of observations in 2000 in different age groups

Statistical methods

Handling of the sampling design and missing data

In order to handle sampling uncertainty in the survey data, we generated 36 weighted bootstrapped data sets33 using the poststratification weights to account for oversampling and non-response in the Health 2000 Survey. As the interval between measures was 11 years, we report the results based on 11 year age groups, starting at age 19 years, which allows us to monitor changes in the same birth cohorts or age groups over time. As walking disability is relatively rare under age 52 years, we report projections only in age groups aged 52 years and over.

Non-response in the Health 2011 Survey and item non-response in the Health 2000 Survey were handled by multiple imputation (MI) separately in each bootstrap sample.34 The MI was based on the chained equations, using the classification and regression trees (CART) implemented in the mice package of the R software35 with all results based on the MI. We included all analysis variables and auxiliary variables in the imputation model, which was also applied in the creation of the projections. The CART method accounts for possible non-linear relationships and interactions between the variables. For young adults, the data did not include walking difficulty or the strenuousness of leisure time physical activity variables, which were multiply imputed.

Projections

The projections were generated using MI sequentially for the years 2022, 2033 and 2044 following the MI of the Health 2000 and 2011 data sets. The MI techniques allow for proper handling of the uncertainty inherent in all projections.

Individual follow-up data allowed for an estimation of the incidence rates of a multistate model.26 The three possible states were mobile, disabled or dead, which was the absorbing state. It was also possible to move from state disabled to mobile. Our primary assumption was that risk factors and mobility limitations will change during each 11-year projection period with the same transition probabilities as between 2000 and 2011 (=null scenario) while accounting for the parameter uncertainty. The point estimate of each projection was calculated as the average of the 36-point estimates based on the imputed bootstrap data sets. The 95% prediction interval (PI) limits were calculated using the normal approximation and the SD of these 36-point estimates.

We compared the null scenario with scenarios based on modifications for smoking, physical inactivity and excess weight. First, only one risk factor was changed at a time. In the ‘Smoking50%’ scenario, 50% of daily smokers were randomly chosen and moved to the non-smoker category, and this random assignment was conducted separately in each bootstrap sample. Correspondingly, in the ‘Physical inactivity 50%’ scenario, 50% of persons in the lowest activity category were moved to the higher activity category. In the ‘BMI50%’ scenario, all BMI values above 25 were replaced by the average of the BMI value and 25, thus reducing half of the excess weight. For example, for an individual with BMI 32, the BMI value was replaced by the value 28.5. The scenarios were chosen to illustrate the importance of prevention by demonstrating theoretical contribution of risk factors to the burden of immobility. Second, all risk factors were modified at the same time either by 50% or 100%, respectively. In the ‘All risk factors, 50%’ scenario, each risk factor was changed by 50% as described above. In the ‘All risk factors, 100%’, all individuals who smoked were moved to the non-smoking category and all physically inactive individuals to the higher category, and all BMI values above 25 were moved to the value of 25. The risk factor values were modified at the years 2011, 2022 and 2033 before projecting the next projection point. After the risk factors had been modified, they were assumed to follow the same transition probabilities as in the null scenario, over the next 11-year period. In all scenarios, including the null scenario, mortality was assumed to decrease at the same pace as in the periods 1987–1991 and 2007–2011. The contrasts between the null scenario and the other scenarios were calculated as the differences in each of the 36 imputed datasets. Further details on statistical methods can be found in the online supplementary material.

Results

Transition probabilities 2000–2011

Severe mobility limitations

Mortality was much higher among those who had severe mobility limitations in 2000 compared with those without limitations at baseline, except in the oldest age group (figure 1). The probability of moving from severe (major difficulties/not able) to mild (minor difficulties or not at all) mobility limitations decreased with age (figure 1A). Almost half of those aged 62 years or less at baseline who survived until 2011 recovered from severe mobility limitations, but only a third of the older persons recovered. However, the incidence of severe mobility limitations increased up to the age group 74–84 years: only a few per cent of the younger survivors but more than a third of persons aged 74–84 years developed severe mobility limitations during the 11-year follow-up (figure 1B).

Figure 1

Transition probabilities among persons (A) with and (B) without severe mobility limitations at baseline by age group at baseline.* (A) Persons with severe mobility limitations at baseline. (B) Persons without severe mobility limitations at baseline. *Numbers of participants are within the braces.

Obesity

No mortality differences between initially obese and non-obese persons were found in any age group (table 1). Up to the age group 41–51 years, the transitions from obesity to no obesity and vice versa were almost as likely, but in the older age groups, loss of weight was more likely.

Smoking

Mortality among baseline smokers was twofold to threefold compared with non-smokers in the age groups below 74 years (table 1). The smoking cessation probabilities ranged from approximately 30% under age 52 years to more than 50% in age groups 63–84 years. The relapse probabilities were at most 3%–5% in the age groups below 52 years.

Physical inactivity

Physical inactivity was associated with increased mortality in the age groups 52+ years (table 1). Among those physically active persons at baseline, the probability to become inactive was about 20% between ages 30 years and 84 years, and the corresponding probability of dying during the follow-up increased from 2% to 53%. Almost half of the physically inactive persons aged 30–62 years at baseline became physically active, but in the older age groups, the probabilities decreased and the transition to death increased rapidly.

Projections 2011–2044

Number of persons with severe mobility limitations

Under the null scenario, the number of persons with severe mobility limitations is projected to double from year 2011 to 2044 (from 197 000 to 402 000 persons, 104% increase) in the Finnish population aged 52 years or over (figure 2, table 2). This growth rate is substantially higher than the increase in the population size, resulting in an increasing proportion of persons with disability.

Figure 2

The effect of different scenarios* on the number of persons with severe mobility limitations in years 2022, 2033 and 2044. Point estimates and prediction intervals. *Baseline projection ‘Null’; All50%, all risk factors, 50 %; All100%, all risk factors, 100 %; PA,  p hysical inactivity. †Averages of the 36-point estimates based on the imputed data sets. ‡Ninety-five per cent prediction intervals are based on the SD of the 36 point estimates.

Table 2

Effects of different scenarios on the number of persons with severe mobility limitations, prevalence of mobility limitations and population size in years 2022, 2033 and 2044, in the Finnish population aged 52+ years with 95% prediction intervals (PIs)

Compared with the null scenario, in the ‘BMI50%’, ‘All risk factors, 50%’ and ‘All risk factors, 100%’ scenarios, the number of persons with severe mobility limitations appear to decrease in all projection years (table 2). This is seen in the age groups 52–84 years (online supplementary table S9). In the oldest age group (aged 85 or above), the number of cases will increase multifold in all scenarios, mainly due to the rapid increase in the population size, but the small sample size hinders interpretation of the projections. The ‘BMI50%’ scenario appear to reduce the number of persons with severe mobility limitations much more than the other single risk factor scenarios: the number of persons with severe mobility limitations would only increase by 62% during the period 2011–2044. If half of all three risk factors could be removed (scenario ‘All risk factors, 50%’), the increase would be slightly lower, 59%. Even if excess weight, smoking and physical inactivity could all be completely eliminated (scenario ‘All risk factors, 100%’), the increase would be 46%.

Prevalence of severe mobility limitations

The projected prevalence of severe mobility limitations in the age group 52 years or older will increase from 10% in 2011 to 16% in 2044, due to the rapidly growing size of the older population (table 2, figure 3). The projections based on the ‘BMI50%’ scenario suggest that the prevalence will be about 20% lower than in the null scenario. The difference will be greatest among persons aged 74–84 years (online supplementary table S9). The ‘Smoking50%’ and ‘Physical inactivity50%’ scenarios appear to have no significant impact on the prevalence of mobility limitations. The ‘All risk factors, 100%’ scenario results in the lowest projected prevalence of mobility limitations. The prevalence changes paralleled the changes in the number of persons with mobility limitations, as the population size was not markedly affected by any of the scenarios.

Figure 3

The effect of different scenarios* on the prevalence of severe mobility limitations in years 2022, 2033 and 2044, by age group. *Baseline projection ‘Null’. All 50%, all risk factors, 50%; All100%, all risk factors, 100%; PA, physical inactivity; PI, prediction interval.

Discussion

Our projections demonstrate that the number of persons with severe mobility limitations is likely to double from 200 000 persons in 2011 to 400 000 persons by 2044 in Finland and the prevalence of severe mobility limitations to increase from 10% to 16.4% if the risk factors and mobility limitations continue developing similarly as during the period from 2000 to 2011. This estimated rapid increase has potential to be markedly slowed down by eliminating half of the excess weight: both the number of persons and the prevalence of severe mobility limitations would reduce by one-fifth. Changes in smoking and physical activity had little effect on the projections. This can be attributed to the selection by the higher mortality of smokers and physically inactive individuals. Moreover, there is a two-way association between obesity and physical activity,36 and BMI, which is measured more accurately, may partly over-rule the effect of physical activity. Due to population ageing, the number of persons with severe mobility limitations is expected to increase even if all the most essential behavioural risk factors were to be eliminated.

Our projection on the increase of the number of persons with mobility limitations is well in line with the calculations by Smith et al.23 Projections concerning mobility limitations are rare, even though mobility restrictions are common in old age and precede difficulties in activities of daily living and need of care.3 4 37 We chose mobility limitations as an outcome since its risk factors are well established. Our assumption on eliminating all risk factors is unrealistic, but our results suggest that reducing excess weight by half would yield more than two-thirds of the positive outcome generated by completely eliminating all three risk factors.

Several researchers have raised the question of the deteriorating health of the younger generations, which in turn affects the future disability rates.7 10 12 In projections spanning several decades into the future, it is important to project also the risk factor distributions of the presently young population so as to incorporate their influence on disability. Previous studies25 have mainly focused on older people, while our analyses included all persons aged 19 years or over.

Our method uses longitudinal data, which allow us to take account of the dynamic transitions in the major risk factors and mobility limitations, as well as the interactions both between the predictors and with education. Our projections also take into account the future development of the educational structure and allow for differential survival in the educational groups. A strong association between education and health-related outcomes, including disability,38–40 has been shown, and changes in the educational structure of the population contribute to past9 and future14 21 25 estimates of disability. Another strength in our study is that we accounted for the uncertainty inherent in all predictions. We report the uncertainty using PIs, which accommodate the different sources of uncertainty. The more distant future we are predicting, the wider these intervals get.

There are a few limitations in the present study. Our approach considers only three individual risk factors affecting disability rates. We were not able to cover other individual risk factors, for example, binge drinking, nor environmental adaptations, such as assistive device use, accessible housing and public transportation or various social factors, which may also have a large impact on future disability rates.41 42 Our projections did not account for immigration, which will increase the number of people with disabilities later, as those immigrants have aged. We assumed a fixed decreasing trend in mortality, which might not continue in the future. Mortality being higher or lower would affect the projections. The moderate response rate can cause selection bias. However, the response rate at baseline was high (92%), allowing for a quite accurate imputation of the missing outcome and risk factor values in 2011.31 As our results were based on an observational study, the projections based on different scenarios do not necessarily reflect the causal effects of risk factor changes, which would require incorporation of randomised controlled trials and evidence synthesis.43 44

In conclusion, the number of people with severe mobility limitations is likely to double in the coming decades in Finland. This increase is mainly due to the ageing of the population and will lead to substantial societal and economic consequences, such as increasing care needs and long-term care expenditure and changing demands for the housing industry.23 Nevertheless, by reducing excess weight, the rapid increase could be slowed significantly. Designing and implementing strategies and policies to promote healthy weight and weight reduction as well as the health and well-being of people with mobility limitations should, therefore, be national priorities.

What is already known on this subject

  • Increasing trends in mobility limitations have raised concern about future financing of services for older people.

  • Projections on future development of mobility limitations are rare and have not investigated the impact of changes in modifiable risk factors.

What this study adds

  • By applying multistate modelling, this study accounts for individual changes in risk factors over time, and how these factors may influence mortality and mobility; the different sources of uncertainty in projections are also accounted for.

  • Mobility limitations are projected to increase considerably due to population ageing, but reductions of excess weight would markedly slow the rapid increase.

  • Healthy public policy should prioritise allocating resources to maintaining functional ability and to provision of adequate services for persons with disabilities.

References

Footnotes

  • TH and PS contributed equally.

  • Contributors TH and PS contributed equally in the manuscript preparation. TH, PS and SK participated in concept and design, and all authors participated in interpretation, writing and approved the final version.

  • Funding This work was supported by the Academy of Finland (grant numbers 266251, 286294, 294154 and 307907).

  • Competing interests None declared.

  • Patient consent Not required.

  • Ethics approval Both surveys were approved by the Ethics Committee of the Hospital District of Helsinki and Uusimaa.

  • Provenance and peer review Not commissioned; externally peer reviewed.