Cancers in Australia in 2010 attributable to insufficient physical activity

Abstract

Objectives

To estimate the proportion and numbers of cancers occurring in Australia in 2010 attributable to insufficient levels of physical activity.

Methods

We estimated the population attributable fraction (PAF) of cancers causally associated with insufficient physical activity (colon, post-menopausal breast and endometrium) using standard formulae incorporating prevalence of insufficient physical activity (<60 minutes at least 5 days/week), relative risks associated with physical activity and cancer incidence. We also estimated the proportion change in cancer incidence (potential impact fraction [PIF]) that may have occurred assuming that everyone with insufficient activity levels increased their exercise by 30 minutes/week.

Results

An estimated 1,814 cases of colon, post-menopausal breast and endometrial cancer were attributable to insufficient levels of physical activity: 707 (6.5%) colon; 971 (7.8%) post-menopausal breast; and 136 (6.0%) endometrial cancers. If those exercising below the recommended level had increased their activity level by 30 minutes/week, we estimate 314 fewer cancers (17% of those attributable to insufficient physical activity) would have occurred in 2010.

Conclusions

More than 1,500 cancers were attributable to insufficient levels of physical activity in the Australian population.

Implications

Increasing the proportion of Australians who exercise could reduce the incidence of several common cancers.

Regular physical activity is important for optimal health and significant benefits occur from even modest amounts of physical activity,¹ including lower cancer rates.² There is consistent evidence from a systematic review and meta-analysis of cohort studies that physical activity is associated with a reduced risk of colon cancer.³ Research has also consistently demonstrated links between physical activity and reduced risk of post-menopausal breast cancer⁴ and endometrial cancer.⁵ The World Cancer Research Fund (WCRF) has concluded that there is “convincing” or “probable” evidence that insufficient physical activity causes cancers of the colon, post-menopausal breast and endometrium.⁶

The most recent Australian guidelines for physical activity and sedentary behaviour were released in February 2014.⁷ They recommend that adults perform at least 150 minutes of moderate intensity physical activity or 75 minutes of vigorous intensity physical activity per week to help improve blood pressure, cholesterol, heart health and muscle and bone strength. This should be increased to 300 minutes of moderate intensity physical activity or 150 minutes of vigorous intensity physical activity per week to reap greater health benefits and help to prevent cancer and unhealthy weight gain.⁷

The National Guidelines defines 60 minutes of moderate intensity physical activity on most days of the week (assumed 5 days) as a sufficient level to help prevent cancer. We aimed to estimate the fraction and number of cancers of the colon, post-menopausal breast and endometrium arising in the Australian population in 2010 that were attributable to failing to meet this target. We assumed that lower levels of physical activity conferred some benefit, but less than optimum.

Methods

Physical activities are often categorised by domain, e.g. occupational or recreational, as well as by type, frequency, duration and intensity. Different types of activity are commonly equated through metabolic equivalents (METs). One MET is considered to represent resting energy expenditure, about 3.5 mL O₂/kg/min when measured in terms of oxygen consumption. Because any form of activity requires increased oxygen consumption, activities can be quantified in terms of multiples of this resting oxygen consumption (e.g. an activity that requires four times the oxygen consumption of rest would be defined as 4 METS). Moderate activity consumes about 3–6 METS,⁸ depending upon fitness level.⁹ We followed the approach of a UK PAF project¹⁰ and assumed, conservatively, that moderate activity is equivalent to 6 METS. It follows that exercise of moderate intensity of 60 minutes duration consumes 6 MET-hours. Complying with the National Guidelines equates to 300 minutes or 30 MET-hours per week.

Relative risk estimates

For colon cancer and post-menopausal breast cancer, we used relative risks reported in the WCRF Continuous Update for Colorectal Cancer¹¹ and the WCRF Second Expert Panel Report,⁶ respectively. Because the WCRF did not publish a dose-response summary relative risk for endometrial cancer, we sourced the summary relative risk from a dose-response meta-analysis of three cohort studies undertaken by Keum and colleagues¹² (Table 1).

Table 1.

Relative risks for recreational physical activity and colon, breast (post-menopause) and endometrial cancer

Cancer site	Source	Relative risk
Colon	WCRF CUP (2010)11	0.98 (95%CI 0.96–1.00) per 5 MET-hours per week
Breast (post-menopause)	WCRF Second Expert Panel (2007)6	0.97 (95%CI 0.95–0.99) per 7 MET-hours per week
Endometrium	Dose-response meta-analysis Keum et al (2013)12	0.99 (95%CI 0.98–1.00) per 3 MET-hours per week

The increase in risk for a decrease of 1 MET-hour of recreational physical activity per week was estimated by assuming a log-linear relationship between exposure and risk, so that:

Risk per 1 MET-hour per week = ln(1/RR_x)/x

where x is the exposure level (in MET-hours per week) and RR_x the relative risk for x MET-hours per week.

The increases in risk for a decrease of 1 MET-hour of recreational physical activity per week for colon cancer, post-menopausal breast cancer and endometrial cancer were 4.041e-3, 4.351e-3 and 3.35e-3 respectively.

Exposure prevalence estimates

The latent period between physical inactivity and onset of cancer is not known. In the WCRF systematic review of colon cancer, follow-up periods in cohort studies that reported dose-response MET-hours, ranged from 1.6 years to 16 years, with an average of 9.2 years.¹¹ For post-menopausal breast cancer, follow-up periods ranged from 4.7 years to 7.3 years, with an average follow-up of 5.7 years;¹³ and for endometrial cancer, there was an average follow-up of 8.8 years (ranging from 6.6 years to 11.0 years).¹² We used prevalence data from 2001 and cancer incidence data from 2010 to give a nominal latent period of about 10 years. To account for population ageing with time since exposure and the latent period, we used prevalence data for the age category that was 10 years younger than the corresponding cancer incidence age category (for example, cancer incidence in the 25–34 years age group in 2010 was attributed to insufficient physical activity in the 15–24 years age group in 2001).

We used data from the 2001 National Health Survey Confidentialised Unit Record Files¹⁴ to estimate the deficit in physical activity against the recommended guidelines for cancer prevention (300 minutes of moderate activity or about 30 MET-hours per week), by sex and age categories.

The National Health Survey did not directly report the proportion of people undertaking different amounts of activity by age and sex; however, data reported in the National Health Survey Confidentialised Unit Record Files included:

• the mean number of minutes individuals who reported walking, moderate or vigorous exercise spent participating in that activity per fortnight, by sex and age categories; we converted this to mean minutes per week by dividing by 2 (online supplementary file Table S1); and

• the proportion of people (by age and sex categories) doing different combinations of types of exercise (no exercise, walking only, moderate exercise only, vigorous exercise only, walking and moderate, walking and vigorous, moderate and vigorous, walking plus moderate and vigorous).

To estimate the proportion of people exercising at different levels, we first estimated the average total minutes individuals reporting each activity combination spent exercising per week (online supplementary file: Table S2). To do this, we assumed that people who reported two types of activity performed each for half the average duration for each activity and that those who reported all three types of activity performed each for one-third of the average duration. In sensitivity analyses, we assumed that people who performed multiple types of activity in a week spent the average time on each activity type. For example, for people reporting the exercise combination of walking and moderate activity, we simply summed the national average durations for each activity.

We applied a weighting (x2) to vigorous activity to account for the higher intensity level of metabolic activity.⁷ The estimated average total minutes of activity for each combination was then converted to MET-hours per week on the assumption that 60 minutes of moderate activity constitutes 6 MET-hours. We conducted a sensitivity analysis assuming that 60 minutes of moderate activity constitutes 3 MET-hours rather than 6 MET-hours.

For each age, sex and exercise type category, we calculated the deficit from the recommended 30 MET-hours per week (hereafter described as ‘insufficient physical activity’) and categorised this as no deficit, >0-<6 MET-hour deficit, 6-<12 MET-hour deficit, 12-<24 MET-hour deficit, and 24–30 MET-hour deficit. For each sex and age-category, the proportion of people in each MET-hour deficit category was then totalled (Table 2). Relative risks were calculated for each Met-hour deficit category for each age group using the following formula:

Table 2.

Estimated proportion (%) of adult men and women in the Australian population performing physical activity at the given level (against the recommended 30 MET-hours per week) (National Health Survey, 2001).^a

	15–24 yrs	25–34 yrs	35–44 yrs	45–54 yrs	55–64 yrs	65–74 yrs	75+ yrs	15+ years
Males
24–30 MET-hrs deficit	18.1	26.2	32.2	31.1	34.4	29.2	43.0	29.0
12-<24 MET-hrs deficit	37.2	44.7	47.2	54.3	0.0	0.0	0.0	27.4
6-<12 MET-hrs deficit	14.7	24.9	17.2	12.1	64.3	29.5	45.2	34.3
>0-<6 MET-hrs deficit	19.8	4.3	3.4	0.0	1.3	40.7	11.9	5.4
No deficit	10.2	0.0	0.0	2.5	0.0	0.6	0.0	4.0
Females
24–30 MET-hrs deficit	24.1	25.8	30.3	30.1	30.7	37.6	54.6	30.9
12-<24 MET-hrs deficit	62.8	64.6	63.5	65.5	8.1	0.0	29.6	57.5
6-<12 MET-hrs deficit	10.3	7.6	5.0	4.4	58.1	59.9	15.7	10.2
>0-<6 MET-hrs deficit	2.8	2.0	1.3	0.0	2.5	0.6	0.0	1.4
No deficit	0.0	0.0	0.0	0.0	0.6	2.0	0.0	0.0

a: Derived from data from the 2001 National Health Survey Expanded Confidentialised Unit Record File, RADL.¹⁴

Relative Risk =EXP(Increase in log risk per MET-hour deficit per week * deficit in MET-hours per week)

where the deficit in MET-hours per week is the mid-point of the MET-hour deficit categories.

Statistical analysis

The population attributable fraction (PAF) was calculated using the standard formula:¹⁵

where p_x is the proportion of the population in each exercise category x and ERR_x is the excess relative risk (RR-1) for each exercise category x.

To obtain the number of cancers attributable to a deficit in physical activity levels, the PAF was multiplied by the number of incident cancers at each site in 2010¹⁶ for each age and sex category. The number of colon, post-menopausal breast and endometrial cancers attributable to a deficit in physical activity levels was then summed and expressed as a percentage of the total number of all incident cancers (excluding basal cell and squamous cell carcinoma of the skin) recorded among adults aged 25 and over in Australia in 2010.

Potential impact of increasing physical activity levels in the Australian population

We estimated the number of cancers in 2010 that may have been prevented by proposing a hypothetical situation in which all those exercising below the recommended level increased the amount of time they spent exercising at moderate intensity by 30 minutes (3 MET-hours) per week. To do this, we reduced the mid-point of each exercise deficit category by 3 MET-hours (e.g. the mid-point of the 6-<12 MET-hour deficit category was reduced from 9 to 6), and then used the relative risk per MET-hour deficit to estimate the new relative risk for each MET/hr deficit category. We then calculated the potential impact fraction (PIF) using the formula of Barendregt and Veerman:¹⁷

where p_x is the proportion of population in each age and sex category x, RR_x is the relative risk for that category and RR^*_x is the new relative risk for each MET/hr deficit category when the mid-point of each category was reduced by 3 MET-hours.

The PIF is the proportional difference between the observed number of cancers and the number expected under the alternative scenario. Finally, we calculated the number of colon, post-menopausal breast and endometrial cancers that would have not occurred among people aged 25 and over in Australia in 2010 under the alternative scenario.

Results

The estimated proportion of Australian adults meeting the physical activity guidelines for cancer prevention was only 4% for men and less than 1.0% for women (Table 2). The proportion of women doing little or no exercise increased steadily with age, ranging from 24% (15–24 years) to 55% (75+ years). A similar pattern was seen for men, although there appeared to be an increase in activity after retirement age. Of note was the large decline in activity levels between the 15–24 and 25–34 year age groups, particularly for men, suggesting this group might be an appropriate target for intervention activities.

The estimated numbers and proportions of cancers attributed to insufficient physical activity levels are presented in Table 3. In 2010, 27,294 diagnoses of cancers of the colon, post-menopausal breast and endometrium in Australians >25 years, of which we estimated 1,814 (6.6%) were attributable to insufficient physical activity; 707 (6.5%) colon cancer cases, 971 post-menopausal breast cancer cases (6.8% of all breast cancers, 7.8% of post-menopausal breast cancer cases [45+ years]), and 136 (6.0%) endometrial cancer cases. This corresponds to 1.6% of all cancer cases (excluding basal cell carcinoma and squamous cell carcinoma of the skin) in Australian adults aged >25 years in 2010 (0.5% in men and 2.9% in women).

Table 3.

Population attributable fraction (PAF) and estimated number of cancers diagnosed in Australia in 2010 attributable to insufficient levels of physical activity

Age at outcomea	Colon [C18, C19]b			Breast (post-menopausal) [C50]b			Endometrium [C54, C55]b			All cancersc
	PAF	Obs.	Exc.	PAF	Obs.	Exc.	PAF	Obs.	Exc.	Obs.	Exc.
Males
25–34 yrs	5.4	37	2							1,042	2
35–44 yrs	6.9	134	9							2,214	9
45–54 yrs	7.4	469	35							6,632	35
55–64 yrs	7.5	1187	89							16,279	89
65–74 yrs	6.0	1,817	109							19,513	109
75–84 yrs	4.7	1,557	73							14,520	73
85+	6.3	498	32							4,968	32
Total		5,699	349							65,168	349
PAFaw	6.1									PAFaw=	0.5
Females
25–34 yrs	7.4	40	3		258		6.1	19	1	1,401	4
35–44 yrs	7.5	134	10		1,420		6.3	103	7	3,637	17
45–54 yrs	7.8	422	33	8.4	3,385	284	6.5	362	24	7,812	341
55–64 yrs	7.9	908	72	8.5	3,893	330	6.6	721	48	11,042	450
65–74 yrs	6.0	1,346	80	6.4	2,845	182	5.0	556	28	11,073	290
75–84 yrs	6.2	1,522	94	6.6	1,617	107	5.1	355	18	9,819	219
85+	8.4	794	66	9.0	756	68	7.0	139	10	5,166	144
Total		5,166	358		14,174	971		2,255	136	49,950	1,465
PAFaw	6.9			6.8d			6.0			PAFaw=	2.9
				7.8e
Persons
25–34 yrs		77	5		258			19	1	2,443	6
35–44 yrs		268	19		1420			103	7	5,851	26
45–54 yrs		891	68		3,385	284		362	24	14,444	376
55–64 yrs		2,095	161		3,893	330		721	48	27,321	539
65–74 yrs		3,163	189		2,845	182		556	28	30,586	399
75–84 yrs		3,079	167		1,617	107		355	18	24,339	292
85+		1,292	98		756	68		139	10	10,134	176
Total		10,865	707		14,174	971		2,255	136	115,118	1,814
PAFaw	6.5			6.8d			6.0			PAFaw=	1.6
				7.8e

Abbreviations: Obs. = observed cancers in 2010; Exc. = excess cancers in 2010 attributable to insufficient physical activity; PAF = population attributable fraction (expressed as a percentage); PAFaw = age-weighted population attributable fraction (expressed as a percentage).

a: Prevalence data age groups are 10 years younger than cancer incidence age groups assuming a 10 year latent period.

b: International Classification of Diseases Code (ICD-10).

c: Excluding basal cell carcinoma and squamous cell carcinoma of the skin.

d: % of all breast cancers.

e: % of post-menopausal breast cancers (45+ years).

Sensitivity analysis

When the average times per week spent undertaking multiple types of activity were simply summed (rather than proportionally split as in the primary analyses), overall activity levels increased. Consequently, the estimated number of cancer cases attributable to insufficient levels of physical activity dropped to 610 (5.6%) colon cancers, 821 (5.8%) post-menopausal breast cancers (5.8% of all breast cancers, 6.6% of post-menopausal breast cancers [45+ years]) and 113 (5.0%) endometrial cancers.

Our sensitivity analysis, assuming that 60 minutes of moderate activity constitutes 3 MET-hours, resulted in higher estimates of the PAF for colon cancer (8.2%), breast cancer (7.7%) and endometrial cancer (6.8%); with the total number of excess cancer cases increasing from 1,814 to 2,131.

Potential impact of increasing physical activity levels in the Australian population

Reducing the ‘average’ deficit in the activity categories by 3 MET-hours/week would have had a modest impact on cancer incidence. If everyone who did insufficient physical activity had increased their exercise levels by 3 MET-hours/week, about 314 fewer cancers would have been diagnosed (PIF 1.2%). This represents 17% (314/1,814) of all cancers attributable to insufficient physical activity. The proportional impacts were similar across all three cancer types.

Discussion

We estimate that 1,814 cases of colon, post-menopausal breast and endometrial cancer that occurred in Australian adults in 2010 could be attributed to insufficient levels of physical activity, corresponding to PAFs of 6.5% for colon cancer, 7.8% post-menopausal breast cancer (6.8% of all breast cancers) and 6.0% for endometrial cancer.

The estimates of PAF were not especially sensitive to the values of energy expenditure used in the calculations of ‘activity deficit’ with the major influence being the high prevalence of physical inactivity (or activity at less than recommended levels) in the population. These attributable fractions are higher than those reported for the UK¹⁰ for cancers occurring in 2010 (5.3% for colon, 3.4% for breast and 3.8% for endometrium); however, the UK study considered a sufficient level of activity as 15 MET-hrs/week, the UK Department of Health target,¹⁸ which is half the level for cancer prevention recommended by Australian guidelines. Not unexpectedly, the proportion of Australians meeting these cancer prevention guidelines in 2001 was low; however, our dose–response calculations of the PAF took into account any exercise performed under the recommended level. Another factor that may have contributed to the very low prevalence of Australians meeting the physical activity guidelines for cancer prevention in our analysis is the way in which physical activity data are reported. The National Health Survey Confidentialised Unit Record Files only report the proportion of people by age-group and sex who participate in any given activity, and then report the mean number of minutes spent performing that activity. For the purposes of estimating the PAF, it would be more informative to report the proportions of Australians in categories of time spent in moderate and vigorous activity, respectively (e.g. ‘none’, ‘1–29/mins/week’, ‘30–59 mins/week’, ‘60–89 mins/week’, etc). Ideally, future investigations would have access to more complete data to enable better estimates.

Other studies that have reported PAFs for different populations have generally undertaken dichotomous comparisons of active vs. inactive adults, however defined,^19–22 resulting in much higher PAFs. Such estimates are not directly comparable with our findings using dose–response relative risks, which assume that even low levels of physical activity confer some benefits. We contend that this approach generates more plausible estimates of effect than a simple categorical analysis in which it is assumed there is no benefit for failing to meet the guideline, even among people undertaking some physical activity. We also used summary relative risks from cohort studies only, which are less likely to be affected by information biases than risk estimates from case-control studies, but are typically more conservative in their estimates of benefits of physical activity.

These analyses highlight the relative paucity of national data describing durations (as opposed to categories) of different types of physical activity among Australians. We used the available data to estimate the distribution of MET-hours/week of activity within categories of age and sex; however, this required us to make assumptions about certain activity patterns, especially for people engaging in two or more forms of physical activity. Our approach was to reduce their time spent on each separate activity proportionately. It is possible that this approach was overly conservative, and so we performed sensitivity analyses allowing a more liberal summation of activity levels. Even so, this made little substantive difference to the estimates: most Australians perform insufficient activity, and the estimates of the cancer burden changed little regardless of how activity levels were summed.

We were limited to considering recreational physical activity only, and were unable to consider the potential adverse effects of sedentary behaviour separately. Sedentary behaviour may be particularly important for colon, endometrial and breast cancer risk,^23,24 with a recent meta-analysis reporting increased risks associated with high levels of total sitting time.²⁵

Notably, our calculations of the fraction of cancers attributable to physical inactivity should be independent of the effects of overweight/obesity. We used relative risk estimates that were adjusted for the potentially confounding effects of other exposures, including overweight/obesity, although it is possible that some residual confounding remains.

To avoid potentially subjective assessments of causality, we restricted our analyses to cancers causally associated with physical inactivity as determined by independent agencies (in this case WCRF) that have undertaken systematic and continual review of the evidence.^11,26–32 While we are aware of a growing literature describing possible associations with other cancers (e.g. oesophageal,³³ bladder,³⁴ gastric,³⁵ lung,³⁶ prostate³⁷ and pancreatic³⁸ cancer), the evidence to date has not been sufficient for WCRF or IARC to make a causal judgement, and so these cancers were not included in our analyses.

Maintaining a sufficient level of activity potentially offers people an opportunity to reduce their risk of colon, breast and endometrial cancers in addition to the numerous other chronic diseases associated with low physical activity. There are a number of biologically plausible pathways that may underlie a protective effect of higher levels of activity.

For colon cancer, postulated mechanisms include decreased gastrointestinal transit time (reducing exposure to carcinogens),³⁹ changes in serum levels of insulin and insulin-like growth factors (IGFs),⁴⁰ and alterations in the level of prostaglandin E2.²⁰ Regular physical activity significantly lowers insulin levels and enhances insulin sensitivity, independently of body mass index (BMI) – a measure of body fatness calculated as weight in kilograms divided by the square of height in meters (kg/m²)^40,41 – and insulin increases the bioactivity of insulin-like growth factor I.^42,43 In laboratory studies and animal models, insulin is a growth factor for colon mucosal cells,⁴⁴ and physical activity causes a decrease in prostaglandin E2, which has been shown to stimulate colon cell proliferation in cell culture.⁴⁵ Indirect evidence for a role of prostaglandin E2 derives from the observation that aspirin and nonsteroidal anti-inflammatory drugs – inhibitors of prostaglandin synthesis – reduce risk of colon cancer.⁴⁶

Proposed mechanisms for the benefits of physical activity on breast and endometrial cancer development include alterations in endogenous sex hormone levels (oestrogen, progesterone and androgens), insulin-mediated pathways, and maintenance of energy balance.^20,47 Regular physical activity has been shown to lower the levels of biologically available oestrogens, progesterone, and androgens^48,49 and to increase levels of circulating sex hormone binding protein.⁵⁰ Elevated serum levels of androgens and oestrogens in both pre- and post-menopausal women are associated with increased endometrial cancer risk,⁵¹ and elevated circulating oestrogens in post-menopausal women are associated with increased breast cancer risk.⁵² Regular physical activity also helps prevent or reduce obesity with consequent improvement in the metabolic profile (endogenous hormone and growth factor levels),⁴⁷ although it can be difficult to disentangle the independent effects of increasing physical activity and weight loss.

Sequential National Health Surveys have indicated that the proportion of Australians who are inactive or undertake only low levels of physical activity has declined from 73% in 2007–08 to 68% in 2011–12.⁵³ The 2011–12 Survey results indicated that undertaking sufficient levels of physical activity was more likely for people who rated their health as “excellent”, who lived in areas of least disadvantage, and who had higher incomes and higher levels of education.⁵³ It is possible that those Australians who would stand to benefit most by engaging in higher levels of physical activity are not becoming more active, and this could have implications for future levels of cancer (and other chronic diseases) in the Australian community. While evidence on the effectiveness of workplace physical activity interventions is equivocal,⁵⁴ community-wide health education campaigns and school-based interventions have been shown to be effective,⁵⁵ supporting the role of multi-component interventions to successfully increase physical activity levels.

PAF Project

Chief Investigators: David C. Whiteman, Penelope M. Webb, Adele C. Green, Rachel E. Neale, Lin Fritschi

Associate Investigators: Louise F. Wilson, Catherine M. Olsen, Christina M. Nagle, Nirmala Pandeya, Susan J. Jordan, Annika Antonsson, Bradley J. Kendall, Torukiri I. Ibiebele, Maria Celia B. Hughes, Kyoko Miura, Susan Peters, Renee N. Carey

Advisers: Christopher J. Bain, D. Max Parkin

Supporting Information

Additional supporting information may be found in the online version of this article: