Introduction
There is ample empirical evidence that physical activity improves (current) well-being and reduces future risks of disease. Thus, physical activity presently exerted has both an immediate welfare enhancing effect as well as a precautionary investment effect. While the immediate effect on welfare is strictly private, the reduction in risks of disease influences both private and public domains. It has been shown that physical activity improves health and promotes longevity.1–4 Physical activity reduces the risk of (1) cardiovascular disease,5 6 (2) hypertension and stroke,7–9 (3) type 2 diabetes,10–16 (4) colon cancer,17–20 (5) breast cancer,21–23 (6) osteoporosis24–26 and (7) depression and anxiety.27–30
In spite of the large and well established health effects and general well-being effects of physical activity, a significant proportions of adult population do not exert any physical activity at all or are irregularly physically active (we will refer to this as being insufficiently physically activity). The WHO reports that 25% of the adult global population is insufficiently physically active. The proportion of the adult Swedish population that partake in physical activity regularly has risen over the last three decades, from about 30% in the early 1980s to more than 60% in the early 2010s. The proportion of women who are regularly physically active was somewhat below that of men in the early 1980s. Since then, the proportion of active women has risen more than that of men, and in the early 2010s, the proportion exceeded that of men (Survey of Living Conditions, Statistics Sweden).
The economic burden imposed on societies of physical inactivity, or insufficient physical activity, is considerable. The costs that arise due to the adverse health effects of physical inactivity have been estimated in a small number of studies.31–41 The literature on the economic burden of physical inactivity was summarised in a recent systematic review.42 Studies published in the peer-reviewed international literature have been performed for, inter alia, Australia, Canada, China, Japan, New Zeeland, the UK, the USA and Switzerland. In addition, Ding et al calculated the global economic burden of physical inactivity, divided per country, employing the prevalence approach and extrapolated healthcare unit costs.42 Estimated proportions of national healthcare expenditures attributable to inadequate physical activity were estimated in the range 0.3%–4.6%. Estimated productivity losses varied between studies due to different methodologies.42 To date, no study of the costs of inadequate physical activity among Swedish adults employing Swedish data has been published. This study fills in that gap by presenting such cost estimates for Sweden for the years 2002 and 2016.
Material and methods
The opportunity cost of illness is conceptually comprised of direct and indirect costs. Direct and indirect costs are the values of the resources used and the productivity lost, respectively, due to illness. In this study, the economic costs of inadequate physical activity were estimated using the cost-of-illness methodology combined with a top-down approach and population attributable fractions (PAFs).43 44 Direct cost components were estimated using the prevalence method, while indirect costs components were calculated, following established methodology, as the expected values from future lost production using the human capital approach with cumulative expected losses over a working life time. Direct cost components include hospital-based inpatient and outpatient care and primary care. Prescription drugs were not included. The Swedish Pharmaceutical register (the Swedish National Board of Health and Welfare), comprising information about prescribed and dispenses pharmaceuticals, does not contain any diagnostic information. In the discussion, we provide a comparison of total sales, without any estimates of the amounts attributable to insufficient physical activity, of relevant pharmaceutical for the years 2002 and 2016. Indirect cost components include market and household-based productivity losses due to mortality and work disability (early retirement).
Total numbers of hospitalisations and hospital visits, in Sweden, by disease category were collected from the national register administrated by the Swedish National Board of Health and Welfare. No national registry exists for primary care. Instead, the total number of primary-care visits was estimated using data collected from the administrative registry in the region of Västra Götaland (VGR) (about 1.7 million inhabitants out of a total of 10 million; the figures for VGR were multiplied by 5.9). The estimates of the number of life years and working years lost due to mortality and permanent disability were based on data on mortality and newly granted early retirement by medical condition, age and sex (National Board of Health and Welfare, Causes of Death Register; Försäkringskassan, government agency administrating social security) and life tables for the years 2002 and 2016 (Statistics Sweden).
Unit costs for hospitalisations, per medical condition, were collected from the cost-per-patient database administrated by the Swedish Association of Local Authorities and Regions, while unit costs for hospital visits and visits to primary care were collected from the VGR county council. The number of lost working years was converted into monetary units by multiplying by average annual wages (Statistics Sweden), including the employer contribution to social insurance. Further, a monetary value was estimated for lost household production, based on survey information on household time use patterns (Statistics Sweden). We assumed that permanent disability prevented the retirees from performing any household work. Present values of future market and household production were discounted at 5%.
Available information about physical activity among Swedish adults permits us to calculate PAFs separately for physical inactivity, , and physical irregular activity, , using the formulas:
where and are the shares inactive and irregularly active in the population, and are the (unadjusted) relative risks of disease among inactive and irregularly active. The shares of physically inactive and irregularly active in the Swedish population are collected from the Survey of Living Conditions (Statistics Sweden). We used the relative risks for disease reported by Garrett et al, which are calculated for an adult population (aged >18), together with physical exercise habits for the Swedish population aged 16–64 in 1996/1996 and 2008/2009 in order to calculated population attributable proportions for 2002 and 2016 (there is no reporting for the age group 18–64).34 The attributable proportions calculated in this way are likely to be lower than the ‘true’ proportions, since the relative risks reported by Garrett et al are valid for those aged >18 years of age, and since the relative morbidity risk associated with insufficient physical activity is lower among those above 64 years of age (the absolute risk of illness is increasing with age). The following medical conditions were included: ischaemic heart disease, hypertension, stroke, depression and anxiety, breast cancer, colon cancer, osteoporosis and diabetes type 2. Table 1 summarises exercise habits and the corresponding PAFs for each included disease.
The Survey of Living Conditions (SLC) comprises information as to the intensity and regularity of the respondents’ physical activity on a five-point scale: 1 corresponds to no physical exercise at all; 2 corresponds to physical exercise now and then; 3 corresponds to physical exercise regularly at least once a week; 4 corresponds to physical exercise regularly about twice a week and 5 corresponds to high intensive physical exercise at least twice a week. The relative mortality risks reported by Garrett et al. (2004)34 are computed for three levels of physical exercise regularity: inactive, irregularly active and regularly active. In order to align the Survey-of-Living-Conditions information with the information provided by Garrett et al, we assumed that the five levels in the SLC correspond to the levels in Garrett et al as follows: level 1 in the SLC corresponds to inactive; levels 2 and 3 correspond to irregularly active and levels 4 and 5 correspond to regularly active.