First author and year of publication | Country | Data sources | Design | Sample | PA categories | Types of costs | Covariates adjusted | Major findings | Population-level amount* (%)†, sensitivity/uncertainty analysis | Time frame‡ | Funding/COI |
Anderson 200559 | USA | HeathPartners members survey (1995) linked with administrative healthcare claim data (1996–1999) | Cross-sectional | Members ≥40 years of age (n=4674) | ≥4×30 min/week (yes vs no) | Professional and hospital claims | Age, sex, chronic disease, smoking, BMI | Physical inactivity, overweight and obesity were associated with 23% health plan charges and 27% of national healthcare charges | Statistical sensitivity analysis conducted | 1 year (1997) | HealthPartners Center for Health Promotion/COI statement missing |
Andreyeva 200635 | USA | Health and Retirement Study | Longitudinal | Adults aged 51–61 years and their spouses (n=7338) | Any VPA versus no VPA | Total healthcare cost | Baseline healthcare spending, socio-demographics, chronic health conditions, smoking, alcohol, BMI | PA was associated with a 7.3% reduction in healthcare cost over 2 years | Structural and statistical sensitivity analysis: 13.2% reduction when baseline health was not adjusted | 1 year (2004) | No funding reported/COI statement missing |
Aoyagi 201142 | Japan | Nakanojo Study | Cross-sectional | All willing community residents aged ≥65 years (not severely demented or bedridden; n=5200) | Quartiles based on accelerometer and pedometer Q1=2000 steps/day and 5–10 min/day of activity at >3 METs | Insurance payments for treatment by a doctor or outpatient service of a hospital (no inpatient treatment cost) | Not stated | Increase in PA of 5% of each group by a single ranking leads to 3.7% of total medical expense | 1 year (2009) | Japan Society for the Promotion of Science/no COI declared | |
Brown 200843 | Australia | ALSWH 2001 | Cross-sectional | Women participants aged 50–55 years (n=7004) | High: ≥1200 MET.min/week Moderate: 600–<1200 MET.min/week Low: 240–<600 MET.min/week Very low: 40–<240 MET.min/week None: <40 MET.min/week | Australian Medicare System (outpatient, general practitioner, specialist, and others) | Area of residence, education, smoking, alcohol | Costs were 26% higher in inactive than in moderately active women, and 43% higher in inactive and obese women than in healthy weight, moderately active women | Potential population-level savings: increasing from ‘none’ to ‘low’ without changing BMI: $A39.1 million, with change in BMI: $A47.1 million | 1 year (2001) | Australian Government Department of Health and Ageing/COI statement missing |
Carlson 20145 | USA | NHIS 2004–2010, MEPS 2006–2011 | Cross-sectional | Adults aged ≥21 years (non-pregnant, did not respond unable to do PA; n=51 165) | Active: ≥150 min MVPA/week Insufficiently active: >0–<150 min MVPA/week Inactive: 0 MVPA | Expenditures for all services | Age, sex, race/ethnicity, marital status, census region, area, poverty level, health insurance, education, smoking, BMI | Mean annual difference in inactive adults (compared with active) was US$1437 (29.9%) and in insufficiently active US$713 (15.4%) | Physical inactivity accounted for US$131 (91–172) billion (12.5%), US$117 (76–158) billion (11.1%) after adjusting for BMI; multiple structural and statistical sensitivity analyses, eg, after excluding those with difficulty walking: US$90 (58–122) billion (9.9%); and further adjusted for BMI: $79 (46–112) billion (8.7%) Converted national estimate: $132.9 billion INT | 1 year (2012) | No funding reported/no COI declared |
Chevan 201446 | USA | NHIS 2006–2007, MEPS 2007–2009 | Cross-sectional | Non-disabled adults (did not respond unable to do PA; n=8843) | (1) PA guidelines (strength and/or aerobic PA) (2) Aerobic PA (0; <75; 75–149, 150–299, >300 min/week) | Expenditures for all services | Age, sex, race, income, health status | No significant association between PA and expenditure when adjusted for covariates | 1 year (2012) | No funding reported/No COI declared | |
Cho 201139 | Korea | A study of 250 adults | Cross-sectional | Adults aged ≥ 40 years, selected from community centres (n=250) | Inactive versus acceptable versus active based on questionnaire score | Self-reported healthcare visits and direct expenditure | None | The mean difference between active and inactive persons was US$14.12/month | 1 year (2009) | Korean Government/COI statement missing | |
Codogno 201548 | Brazil | Local municipality health offices healthcare expenditure data | Cross-sectional | Adults randomly selected in five basic healthcare units in Bauru (≥50 years; n=963) | Habitual PA questionnaire score quartiles | Overall healthcare expenditure (all items registered in the medical records) | Age, sex, smoking, blood pressure, BMI | Inverse association between PA and expenditure | PA explained 1% of medicine and 0.7% overall expenditure (statistical sensitivity analysis conducted) | 1 year (2010) | Brazilian Government and Brazilian Ministry of Science and Technology/no COI declared |
Lin 200845 | Taiwan | NHIS 2001, National Health Insurance Research Database 2001 | Cross-sectional | Adults selected from three major regions of Taiwan (n=15 670) | Exercised in the past 2 weeks (yes versus no) | Healthcare claim data (inpatient and outpatient) | Age, sex, ethnicity, marital status, employment status, income, education, | Those who exercised had lower inpatient expenses (2079 vs 3330 NT$) but higher outpatient expenses (9738 vs 9151 NT$) | 1 year (2001) | Taiwan’s National Science Council/COI statement missing | |
Min 201638 | Korea | Korean National Health Insurance Database | Retrospective cohort | 40 to 69-year-old adults who had not changed PA levels during the study period (n=47 290) | Continuously reported exercise that ‘worked up a sweat’ for >1 time/week | Inpatient, outpatient and prescription costs | Age, sex, income, area of residence, smoking, alcohol, BMI (propensity score matching) | Those who were continuously inactive had 11.7% higher medical costs (8.7%–25.3% disease specific) | Multiple years (2005–2010) | National Research Foundation of Korea and Seoul National University Hospital/no COI declared | |
Musich 200344 | Australia | AHMG Insurance Claim Health Risk Appraisal data (1995–1999) | Cross-sectional | AHMG members (n=19 812) | ≤60 min/week (at risk) versus >60 min/week | Claim charges, primarily including inpatient and some ancillary services | None | At risk versus not at risk: $460 versus $A423/year (not statistically significant) | 2 years (1995–1999) | No funding reported/COI statement missing | |
Peeters 201436 | Australia | ALSWH and Medicare system (2001–2010) | Longitudinal | Middle-aged cohort (born 1946–1951) of Australian women (n=5535–6108) | (1) Active (≥40 MET-min/day)/low sitting (<8 hours/day) (2) Active/high sitting (3) Inactive/ low sitting (4) Inactive/high sitting | Total Medicare cost paid by the government and out of pocket | Survey year, marital status, area of residence, education, smoking, BMI, depressive symptoms | Physical inactivity, not prolonged sitting was associated with higher costs ($A94/year) | $A40 million at the national level | 1 year (2010) | Australian Government Department of Health and Ageing and Australian National Health and Medical Research Council/no COI declared |
Pratt 200040 | USA | NMES 1987 | Cross-sectional | Non-pregnant participants aged ≥15 years, without physical limitations (n=20 041) | ≥30 min of MVPA over ≥3 days versus the rest of the sample | Self-reported medical costs confirmed by a survey of medical providers | Age, sex, lifetime smoking status | Lower annual direct medical costs among those who are physically active: US$1019 versus US$1349 | US$29.2 billion, statistical sensitivity analysis conducted Converted national estimate: $103.6 billion INT | 1 year (1987) | No funding reported/COI statement missing |
Pronk 199947 | USA | HeathPartners members survey (1995) linked with administrative healthcare claim data (1995–1996) | Cross-sectional data | Members ≥40 years of age (n=5689) | Number of active days in the prior week | HealthPartners medical claims | Age, sex, race, chronic disease, smoking, BMI | An additional day of PA led to a 4.7% decrease in median medical charges | 1.5 years (1995–1996) | HealthPartners/COI statement missing | |
Wang 200441 | USA | NMES 1987 | Cross-sectional | Non-pregnant adults who reported being downhearted and blue at least a little of the time (n=12 250) | ≥30 min of MVPA over ≥3 days versus the rest of the sample | Medical costs including hospitalisations, physician visits, medication, home care | Age, sex, race, socioeconomic status, area of residence, physical limitations, smoking, body weight | Among those downhearted and blue, physical inactivity was associated with 6.1% of the expenditure (US$133 in 1987 and US$429 in 2003) | Physical inactivity accounted for US$11.8 billioConverted national estimate: $37.2 billion INTn in 1987 (US$38 billion in 2003) among those who were downhearted and blue | 1 year (1987/2003) | No funding reported/COI statement missing |
Wang 200449 | USA | NHIS 1995, MEPS 1996 | Cross-sectional | Non-pregnant adults without physical limitations (n=2472) | ≥5×30 min MPA/week or ≥3×20 min VPA versus the rest of the sample | Self-reported medical costs confirmed by a survey of medical providers | Covariates were not specified, stratified by age groups, sex, smoking status and weight | Active adults and lower prevalence of CVD and lower cost per case of CVD | Physical inactivity accounted for 13.1% of medical expenditure of people with CVD | 1 year (1996) | No funding reported/COI statement missing |
Yang 201137 | Japan | A cohort of the National Health Insurance beneficiaries | Longitudinal | Seniors aged ≥70 years capable of PA, without CVD, cancer, arthritis, and cognitive dysfunction (n=483) | Low: no sports+ no brisk walking + low walking Moderate: no sports + no brisk walking + any walking High: any sports + any brisk walking + any walking | Inpatient and outpatient costs | Age, sex, hypertension, hyperlipidaemia, diabetes, liver or renal disease, smoking, drinking, BMI, physical performance, depressive symptoms, cognitive status | Per capita medical costs: low versus moderate versus high: US$875 versus US$751 versus US$723/month; when adjusted for physical performance: US$827 versus US$711 versus US$702/month (difference driven by inpatient costs) | Statistical sensitivity analysis conducted | 5.5 years (2002–2008) | Japanese Society for Promotion of Science, Japan Atherosclerosis Prevention Fund, Ministry of Health, Labour and Welfare of Japan/COI statement missing |
*Converted national estimate: we inflated the national estimates in local currency units from the year of data to 2013 using the annual consumer prices inflation indicators from the World Bank (http://data.worldbank.org/indicator/FP.CPI.TOTL.ZG) and then converted to purchasing power parity (PPP) international dollars using conversion factors provided by the World Bank (http://data.worldbank.org/indicator/PA.NUS.PPP).
†Interpreted as the total amount of direct healthcare cost that was associated with physical inactivity. % interpreted as the percentage of overall healthcare cost that was spent on diseases that were attributable to physical inactivity.
‡The cost estimate may be based on data from multiple years; however, the time frame here refers to the year for which the estimate is presented. For example, Anderson 200559 averaged annualised charges over a 4-year period (from 1996 to 1999) but presented the estimates in 1997 US$.
A, Australian dollars; AHMG, Australian Health Management Group; ALSWH, Australian Longitudinal Study on Women’s Health; BMI, body mass index; COI, conflict of interest; CVD, cardiovascular disease; INT, International dollars; MEPS, Medical Expenditure Panel Survey; MET, metabolic equivalents; MPA, moderate physical activity; MVPA, moderate-to-vigorous physical activity; NHIS, National Health Interview Survey; NMES, National Medical Expenditure Survey; NT$, New Taiwan dollars; PA, physical activity; Q1, quartile 1; VPA, vigorous physical activity.