Abstract
Introduction
Physical activity (PA) is associated with many health benefits, including decreased risk of cardiovascular disease (CVD), type 2 diabetes (T2D), hypertension, dyslipidaemia, and lower mortality risk (U.S. Department of Health and Human Services, 2008). Active individuals have been reported to have a 30% lower mortality risk during follow-up compared to inactive persons, with inverse associations between PA and mortality risk also being observed in the older adult population (Physical Activity Guidelines Advisory Committee, 2008), and a handful of studies demonstrating this association in very old adults (Ottenbacher et al., 2012; Rizzuto, Orsini, Qiu, Wang, & Fratiglioni, 2012; Yates, Djousse, Kurth, Buring, & Gaziano, 2008). Many studies that have examined the relationship between PA and mortality risk have looked at volume of PA, and there is a paucity of studies that have directly examined the association between frequency of PA and mortality risk (Lee & Skerrett, 2001). Prior studies that have looked at PA frequency have reported that in older adults, participation of PA once or just a few times a week is associated with lower mortality risk compared to being physically inactive (Kushi et al., 1997; Sundquist, Qvist, Sundquist, & Johansson, 2004). Although few studies have examined the association between PA and mortality risk at different stages of adulthood, those that have generally show that PA is associated with lower mortality risk in both younger and older adults (Gulsvik et al., 2012; Kaplan, Seeman, Cohen, Knudsen, & Guralnik, 1987). However, prior research only compared low to high PA, and did not include a very old age group, which is a population that is expected to steadily increase (Federal Interagency Forum on Aging-Related Statistics, 2012). Individuals of various ages are at risk for different health problems, and because PA has differential effects on the various aspects of health, the association between PA and mortality risk may differ by age. In particular, PA is strongly associated with improved physical functioning in very old adults (Stessman, Hammerman-Rozenberg, Cohen, Ein-Mor, & Jacobs, 2009), a factor that is important for this age group. Thus, older adults may be a population that particularly benefit from participating in PA compared to younger age groups. Therefore, the objective of this study was to determine if the association between the frequency of PA and mortality risk differs among middle-aged, old, and very old adults.
Method
Data for the NHANES III was collected on a national sample of 33, 994 individuals’ aged 2 months and older in the United States between 1988 and 1994. A detailed explanation of the data collection methods has been previously reported (National Center for Health Statistics, 1994; National Centre for Health Statistics, Centre for Disease Control and Prevention, 1996). Mortality information for NHANES III survey participants was provided by the National Centre for Health Statistics using probabilistic record matching with death certificate data found in the National Death Index (NCHS Linked Mortality File) through December 31, 2006. All survey participants gave written informed consent prior to participation, and the methods were approved by the National Centre for Health Statistics. Subjects were excluded from this analysis if they were younger than 40 years old, died within the 1st year after they were sampled, or had missing information for physical activity, CVD, T2D, or lipid medications, doctor’s diagnosis of CVD or T2D, body mass index (BMI), self-rated health status, or self-reported mobility impairment. The final sample size was 9,249.
Monthly PA was assessed with questionnaires. Participants were asked if they performed the following activities in the last month: walking 1 mile (1.609 km) without stopping, biking, jogging, swimming, dancing, aerobics, calisthenics, gardening, and weight lifting. They could also list up to four additional activities. Information on the monthly frequency of the activity as well as the intensity of the activity was recorded, and each activity was assigned a metabolic equivalent value. Only activities that were at least a moderate intensity (metabolic equivalent ≥ 3), for example walking or gardening, were included in the analysis. Activity was divided into inactive (0 bouts/week), lightly active (1-2 bouts/week), moderately active (3-4 bouts/week) and very active (5 or more bouts/week). Questionnaires were also used to assess age, sex, education (less than high school, high school, more than high school), ethnicity (white or non-White), smoking status (past smoker, current smoker, never smoker), BMI based on self-reported height and weight, medications (CVD, T2D, lipid), self-reported doctor’s diagnosis of hypertension, heart attack, stroke, congestive heart failure, and T2D, as well as self-reported health status and mobility impairment.
Age was classified into: middle-aged (40-64 years), old (65-79 years), and very old (80+ years). Participants were classified as having CVD if they reported a diagnosis of hypertension, heart attack, stroke, or congestive heart failure, if they were taking any CVD medications or blood pressure medications, or if they had a systolic blood pressure ≥140 mmHg or a diastolic blood pressure ≥90 mmHg. Participants were classified as having T2D if they reported a doctor’s diagnosis of T2D, if they reported taking diabetes medications, or if they had a fasting blood glucose level of ≥7.0 mmol/L. Participants were considered dyslipidaemic if they reported taking any lipid medications, if fasting total cholesterol was ≥6.2 mmol/L, or if fasting triglycerides were ≥2.3 mmol/L. Self-rated health status was reported as excellent, very good, good, fair, or poor. Participants were classified as having a mobility impairment if they reported any difficulty with walking from room to room on one level, rising from an armless chair, or if they reported using any device to help them move, such as a cane, walker, or wheelchair.
One-way analysis of variance was used to assess baseline differences between activity groups with inactive as the referent with Bonferroni-adjusted post hoc analyses. Chi-square tests were used to assess baseline group differences in the categorical variables. Linear and logistic regressions were used to assess trends for differences in subject characteristics across PA levels within each age category and across age categories within each PA level. Hazard ratios for group differences in mortality risk were estimated by Cox proportional hazard analyses for all ages combined in order to explore interactions between PA, age and sex, and for all age categories separately to determine relative mortality risk within an age group. Bonferroni adjustments were used for all post hoc comparisons. The first multivariable model was adjusted for age, sex, education, ethnicity, and smoking status, and the final fully adjusted model included these variables in addition to CVD, T2D, dyslipidaemia, BMI, self-reported health status, and mobility impairment. All analyses were sample weighted to be representative of the U.S. population and were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina). Statistical significance was considered at P < .05.
Results
Subject characteristics are presented in Table 1 stratified by PA level and age category. More women were inactive than men within all age groups. There were no differences in the prevalence of dyslipidaemia across physical activity groups within each age category. There was a trend for higher prevalence of CVD, T2D, and mobility problems with decreasing PA level at all ages (Ptrend < = 0.001). BMI decreased with increasing PA level within all age groups (Ptrend < = 0.001). For all ages, self-rated health tended to be good or excellent with increasing levels of PA, and fair or poor with decreasing levels of PA (Ptrend < = 0.001). There was a trend for higher prevalence of CVD, T2D, mobility impairment, and White ethnicity with increasing age within each PA level (Ptrend < = 0.001). Education level, BMI, and self-rated health declined with age within each PA level (Ptrend < = 0.001).
Baseline Characteristics of 9249 Adults From the NHANES III (1988-1994) Survey Stratified by Physical Activity Level and Age Category.
= significant trend across PA level within age group (P < .05). † = significant trend across age within a PA group (P < .05). BMI = body mass index; CVD = cardiovascular disease; Exc = excellent; PA = physical activity, T2D = type II diabetes. Values are means ± SD or n (%).
During the follow-up of 12.1 ± 4.5 years, there were 3,738 deaths (40%). Results for the first multivariable model and the fully adjusted second model were similar, thus results are only presented for the fully adjusted model. In the analysis with all ages combined, age and PA level were both independently associated with mortality risk (P < .001). There was no significant interaction between sex and activity level (P = .86), however there was a significant age by PA level interaction for mortality risk (P = .04). Thus, when stratified by age category, lightly active (middle-aged: HR 0.71, 95% CI 0.54-0.94; old: 0.77, 0.63-0.93; very old: 0.73, 0.59-0.90; P < .05), moderately active (middle-aged: 0.59, 0.81-0.86; old: 0.81, 0.67-0.97; very old: 0.73, 0.58-0.92; P < .05), and very active individuals (middle-aged: 0.72, 0.55-0.96; old: 0.73, 0.60-0.89; very old: 0.59, 0.47-0.73; P < .05) had a lower mortality risk compared to inactive within each age category (Figure 1). In the very old adults only, very active also had a lower mortality risk compared to lightly active (0.80, 0.64-0.99; P = .03) and moderately active (0.80, 0.65-0.98; p = .03; Figure 1).
Hazard ratios for mortality risk in 9,249 adults from the NHANES III (1988-1994) survey stratified by age and physical activity level. Group comparisons were made between all physical activity levels within each age group.
Discussion
We observed that in a nationally representative U.S. sample of middle-aged, old, and very old adults, participating in moderate-vigorous PA one time a week or more is associated with a lower mortality risk compared to being completely inactive. However, this is the first study to show that very old adults who participate in moderate-vigorous PA five or more times per week have a significantly lower mortality risk compared to very old adults who are physically active but participate in a lower frequency of PA. Thus, in agreement with the previous physical activity guidelines, these results suggest that all adults, and in particular very old adults should be encouraged to participate in physical activity five or more times/week.
Despite the known health benefits of regularly participating in PA, in 2008, ~25% of the overall U.S. adult population and ~32% of U.S. older adult population participated in no leisure-time PA (Centres for Disease Control and Prevention, Department of Health and Human Services, 2010). Physical inactivity is a known contributor to many chronic diseases in older Americans including vascular disease, T2D and some forms of cancer (Agency for Healthcare Research and Quality, Department of Health and Human Services, 2002), and medical costs are higher for inactive adults than active adults (Garrett, Brasure, Schmitz, Schultz, & Huber, 2004). Accordingly, adults who meet the 2008 Physical Activity Guidelines have a lower mortality risk compared to adults who did not meet the guidelines (Schoenborn & Stommel, 2011). However, one study reported that older adults who were physically active even once per week were at a lower mortality risk compared to inactive older adults (Sundquist et al., 2004). Our study confirms these observations by demonstrating that even very low levels of PA are associated with lower mortality risk in all adults, but extends these findings in that engagement in higher frequencies of PA is associated with even lower mortality risk in adults 80 years or older. Beneficial effects of PA on mortality risk in very old adults have been observed in several other studies (Aijo & Parkatti, 2011; Paganini-Hill, Kawas, & Corrada, 2011; Stessman et al., 2009). Among individuals without heart disease, the volume of PA is associated with a greater benefit on mortality risk for adult’s ≥ 75 years, compared to those < 75 years (Bembom, van der Laan, Haight, & Tager, 2009). Due to the natural deterioration of health that is associated with aging, older adults are much more likely to be at risk for health problems which may be preventable by PA (i.e., vascular disease, diabetes, respiratory disease, and cancer). Together, this suggests that being physically active in late adulthood may be associated with greater declines in mortality risk than being physically active in early or middle adulthood.
Although the present analysis was able to assess PA levels using intensity and frequency of PA, information on duration of activity was not available. The volume of PA is most commonly used to prescribe PA (Physical Activity Guidelines Advisory Committee, 2008), and the most current guidelines no longer give a recommended frequency for PA. In the present study, a frequency of PA of five times a week or greater was associated with a 27% lower mortality risk for very old adults. Although we cannot discount that the beneficial effect of a higher PA frequency is due to a higher volume of PA, there is evidence that elderly men who expend the recommended ≥ 1000 kcal/week in just 1 or 2 days (“weekend warriors”) still have a higher mortality risk compared to those who accumulate ≥ 1000 kcal/week but are regularly active throughout the week (Lee, Sesso, Oguma, & Paffenbarger, 2004). In addition, it has been reported that even one acute bout of exercise can significantly reduce blood pressure up to 16 hr, blood glucose for several days, and trigylcerides up to 72 hr postexercise, with the effects usually being greatest in high risk or diseased populations (Thompson et al., 2001). Thus, a higher frequency of PA may make it more likely that the individual chronically benefits from the acute effects of PA. Therefore, at least in older adults, frequency of PA perhaps should be considered when prescribing PA.
Despite findings that adults ≥ 80 years can benefit from PA, there is evidence that very old adults participate in less PA than middle-aged and old adults (Publications Statistics, Department of Health, 2000). Older adults may have additional barriers that prevent them from exercising, such as a greater number of comorbidities (Belza et al., 2004), fear of injury (Prohaska, Peters, & Warren, 2000), and they may not agree with some of the PA recommendations (Sims, Hill, Davidson, Gunn, & Huang, 2007). Thus, the issue of addressing the higher rates of physical inactivity in this age group may be more complex than other populations.
Several limitations of this analysis warrant mention. This study used prospective observational data and therefore no causal relationships can be determined. It may be that adults who did not participate in any activity were in poorer health to begin with, which was evident by the higher prevalence of comorbidities, lower self-reported health, and higher mobility impairment in the inactive groups at any age. However, it is important to note that even after accounting for the effect of poor health by adjusting for common metabolic health disorders (CVD, T2D, and dyslipidaemia) as well as self-rated health status and mobility impairment, the significant association between PA and a decreased mortality risk was still present. We also limited analyses to only those individuals who were still alive at least 1 year after their information was collected to minimize potential bias of poor health leading to decreased PA. Nevertheless, our data does not account for differences in the severity of these conditions and may have contributed to the differences observed here. Further, given the large age range of our cohort, we cannot exclude the possibility that differences in education, norms regarding behaviors or other unknown factors may have influenced the results here. We have attempted to adjust for some of these factors by adjusting for education and ethnicity, but cannot ignore that there may be secular or cohort differences in how these factors influenced the associations observed in the present study. Another limitation was that PA was only evaluated for the month prior to the survey interview, and lifetime PA patterns were not known. However, there is evidence to suggest that there is no modifying effect of past PA on the association between current activity and mortality risk in adults (Bembom et al., 2009). Still, we cannot preclude that individuals may have altered their PA patterns after our assessment. However, these changes in PA patterns would likely result in an underestimation in the importance of PA observed in this study.
In conclusion, across adulthood engaging in any number of moderate to vigorous PA bouts per week is associated with significantly lower mortality risk compared to being completely inactive. However, in adults 80 years and older, engaging in PA five or more times/week was associated with an even lower mortality risk. Therefore, although PA may be beneficial for all ages, the association between PA frequency and mortality risk is strongest in very old adults. Thus, in accordance with the previous PA guidelines, all adults, and in particular very old adults may benefit from participating in PA five or more times a week.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Centers for Disease Control and Prevention (CDC) funded and conducted the NHANES III study with mortality follow-up and made the data publicly available.
Sponsor’s Role: The CDC had no role in this study. The authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.