Gender Differences in Trajectories of Physical Activity Among Older Americans With Diabetes

Abstract

Objective: The primary objective of this study was to examine gender differences in engagement in physical activity over time among older U.S. adults with diabetes. Method: Using data from the Health and Retirement Study, we investigated physical activity between 2004 and 2010 among 1,857 adults aged 65 years and above with diabetes. Results: Less than half of adults were physically active at baseline. The probability of physical activity declined over the 6-year period, with no significant gender variation in the effect of time. Because the odds of physical activity were lower for women at baseline and the effect of time did not vary by gender, the trajectory of physical activity was less favorable for women than men. Discussion: The women in this cohort of older Americans started and remained less active than their male counterparts. Investigations covering a larger portion of the life course and those examining the impact of life events and transitions on physical activity among adults with diabetes are needed.

Keywords

diabetes gender health behavior longitudinal physical activity trajectories

Introduction

In the United States, approximately one in four adults aged 65 years and above has diabetes (Centers for Disease Control and Prevention [CDC], 2014). Diabetes has been linked with a variety of adverse health outcomes, including heart disease, stroke, peripheral artery disease, neuropathy, kidney disease (Engelgau et al., 2004; Fowler, 2008), and suboptimal cognitive health (Exalto, Whitmer, Kappele, & Biessels, 2012). It is, in fact, one of the leading causes of death among older adults (Federal Interagency Forum on Aging-Related Statistics, 2012).

Regular physical activity is a critically important self-management behavior for those with diabetes, as it promotes glycemic control and enhances insulin action (Colberg et al., 2010). People with type 2 diabetes are encouraged to engage in a minimum of 150 min of moderate to vigorous aerobic exercise over the course of 3 or more days each week, with at most 2 days between exercise sessions (Colberg et al., 2010). Recent national estimates indicate, however, that the majority of older U.S. adults with diabetes do not engage in optimal levels of physical activity and are less likely to engage in recommended levels of physical activity than their peers without the condition (Zhao, Ford, Li, & Balluz, 2011).

Although the number of nationally representative studies of physical activity among older Americans with diabetes is limited, existing research suggests that older women with diabetes are less likely to engage in physical activity than older men. Using data from the 2007 Behavioral Risk Factor Surveillance System, Zhao et al. (2011) examined correlates of meeting the American Diabetes Association’s (ADA) recommendation for physical activity. After controlling for differences in such factors as age, education, race and ethnicity, disability, and heart disease, the odds of women meeting guidelines were nearly one-quarter lower than for men. Importantly, women’s lower level of engagement in physical activity is not without consequence. Recent research suggests, for example, that midlife and older women with diabetes are at greater risk of functional health problems than their male counterparts (Chiu & Wray, 2011b). This appears to be due, in part, to women’s lower level of engagement in physical activity (Chiu & Wray, 2011a).

Although the reasons for the observed gender difference in physical activity among older adults with diabetes are unclear, Bird and Rieker’s (2008) model of constrained choice offers insight. This model suggests that while individuals can and do make choices about various health-related behaviors (e.g., to exercise regularly or not), those choices are influenced by a number of “constraints” (e.g., the social norms, values, and beliefs to which one is exposed and the responsibilities that accompany various social roles; Bird & Rieker, 2008). Of particular relevance to this investigation, Bird and Rieker (2008) argue that the constraints that men and women face may not be the same and/or the impact of a particular constraint (e.g., neighborhood disorder) on health-related choices may vary by gender.

While there are undoubtedly many “constraints” that contribute to women’s lower level of engagement in physical activity, research suggests that women’s caregiving and familial responsibilities may play a major role (Vrazel, Saunders, & Wilcox, 2008; Wilcox, Oberrecht, Bopp, Kammermann, & McElmurray, 2005). Moreover, over the course of their lives, many older women were exposed to negative messages about women and girls’ participation in sports (Lutter, 1994) and likely had limited exposure to physically active role models who could help counter these negative messages (Vrazel et al., 2008; Wilcox et al., 2005). Thus, being physically active is unlikely to have been a normative behavior for many of today’s older women. Other factors that may contribute to lower levels of physical activity among women are concerns about personal safety (Bird & Rieker, 2008; Conn, 1998) and gender differences in health (Shaw, Liang, Krause, Gallant, & McGeever, 2010), including women’s heightened risk of functional health problems and depression (Chiu & Wray, 2011b; Crimmins, Kim, & Solé-Auró, 2011; Roy & Lloyd, 2012).

In summary, we know that older adults with diabetes are less active than those without diabetes, and that older women with the condition are less likely to meet physical activity guidelines than men. We know little, however, about how physical activity among older adults with diabetes changes with time. Compared with their counterparts without the condition, older adults with diabetes are at elevated risk of a number of health problems (e.g., eye disease, heart disease, and neuropathy) that lower the odds of engaging in physical activity (Janevic, McLaughlin, & Connell, 2013). Given the potential for these added challenges to physical activity as well as evidence that suggests that physical activity mitigates the health risks associated with diabetes (Laditka & Laditka, 2015; Palmer, Espino, Dergance, Becho, & Markides, 2012; Stessman & Jacobs, 2014), efforts to understand physical activity in this large and growing segment of the older population are warranted.

Gender patterns in physical activity also warrant additional attention. Although existing research indicates that women with diabetes are less physically active than men, it is possible that the influence of gender on physical activity varies over time. As individuals move through the life course, they transition into and out of social roles that influence their ability to engage in physical activity (Brown, Heesch, & Miller, 2009; Hirvensalo & Lintunen, 2011). Depending on the nature of these transitions and the extent to which they differentially “constrain” or facilitate engagement in physical activity by gender, the gap in men and women’s physical activity behavior may widen or narrow.

Research by Nothwehr and Stump (2000) suggests that the gender gap may widen with time. Using data from the 1992 and 1996 waves of the Health and Retirement Study (HRS), they investigated a range of health practices among adults aged 50 to 62 years with diabetes. Although they observed no statistically significant gender difference in physical activity at baseline, women who were exercising at baseline were more likely to have stopped exercising at follow-up than their male counterparts. This finding needs to be interpreted with caution, however, because the measure of physical activity changed over the study period.

Using data from a panel study of older Americans, we sought to build on the existing body of work by addressing three questions:

Does engagement in physical activity change over time among older adults with diabetes?

Among older adults with diabetes, does engagement in physical activity over time vary by gender?

If a gender difference in engagement in physical activity is evident, does the difference widen with time?

Method

Data Source and Sample

The data utilized in this investigation are from the HRS, a national panel study of adults aged 51 years and over in the United States (Juster & Suzman, 1995). HRS participants are selected using a complex sampling design that involves clustering, stratification, and disproportionate sampling of residents of Florida, those of Black race, and those of Hispanic ethnicity (Heeringa & Connor, 1995).

HRS participants are interviewed every 2 years. Survey interviews cover a range of topics, including engagement in physical activity. Although the study began in 1992, the assessment of physical activity changed in 2004. As a result, we limited our investigation of physical activity to data from the 2004, 2006, 2008, and 2010 waves of data collection. At the time of this analysis, 2010 data were the latest final release data available for analysis.

The analytic sample was restricted to those respondents who (a) indicated that a doctor had ever told them that they had diabetes or high blood sugar as of the 2004 wave of data collection, (b) were aged 65 years and over in 2004, and (c) had non-zero sampling weights (n = 2,147). We excluded those with proxy respondents (n = 221) because they were not administered all relevant measures and those of “other” race and ethnicity due to their small subgroup size (n = 26). The resulting sample included 1,900 respondents. Of those, 43 (2.3%) were missing data for one or more of the variables included in the analysis. Thus, the final analytic sample included 1,857 individuals.

Measures

Time

Between 2004 and 2010, there were a total of four waves of data collection (i.e., 2004, 2006, 2008, and 2010). Using the dates that each individual respondent completed his or her interviews, we created a time variable representing the number of years since 2004. Thus, time equal to 0 corresponds to the 2004 interview, with all subsequent interviews occurring an average of 2.0 to 6.4 years later.

Physical activity

Beginning in 2004, respondents were asked about their level of engagement in light, moderate, and vigorous physical activity. To come as close as possible to the ADA recommendation (i.e., a minimum of 150 min/week of at least moderate-intensity aerobic exercise over the course of 3 or more days per week; Colberg et al., 2010), we used only the moderate and vigorous items.

To assess engagement in vigorous physical activity, participants were asked, “How often do you take part in sports or activities that are vigorous, such as running or jogging, swimming, cycling, aerobics or gym workout, tennis, or digging with a spade or shovel?” They were then asked, “And how often do you take part in sports or activities that are moderately energetic such as, gardening, cleaning the car, walking at a moderate pace, dancing, floor or stretching exercises.” Response options for both items included more than once a week, once a week, one to three times a month, and hardly ever or never. We categorized those who reported engaging in either moderate or vigorous physical activity more than once a week as physically active. All others were categorized as being physically inactive.

Independent variables

The primary independent variable of interest was gender. To help understand any observed gender differences, we also examined a range of demographic and health covariates that have been found to be associated with physical activity in existing research.

Demographic covariates

We examined several baseline demographic characteristics, including age in years (centered on the grand mean), educational level (less than a high school diploma, high school diploma or GED, some college or higher education [referent]), race and ethnicity (Hispanic, non-Hispanic Black, non-Hispanic White [referent]), and quartiles of household wealth (referent = highest quartile of wealth).

Health covariates

Health characteristics assessed at baseline included number of comorbid conditions, level of cognitive functioning, depressive symptoms, and mobility limitations. To capture number of comorbid conditions, a categorical variable was created based on participants’ self-reported history of arthritis, cancer, chronic lung disease, heart disease, hypertension, and stroke (0 = no comorbid conditions [referent group], 1 = one comorbid condition, 2 = two comorbid conditions, 3 = three or more comorbid conditions). Level of cognitive functioning was assessed using a modified version of the Telephone Interview for Cognitive Status (for details, see Herzog & Wallace, 1997; Ofstedal, Fisher, & Herzog, 2005). Scores range from 0 to 35, with higher scores indicating higher cognitive functioning. Depressive symptoms were assessed using an 8-item Center for Epidemiologic Studies Depression Scale (CES-D); individuals obtaining a score of 4 or more (out of a possible eight) were classified as having a clinically significant level of depressive symptoms (Steffick, 2000). The presence of mobility limitations was ascertained using participants’ responses to five items capturing difficulty walking across a room, walking one block, walking several blocks, climbing one flight of stairs without resting, and climbing several flights of stairs without resting. Response options included yes, no, can’t do, and don’t do. For purposes of this investigation, respondents who indicated that they “can’t do” or “don’t do” the task were classified as having difficulty with the task. A total mobility score was obtained by summing the number of tasks with which respondents’ reported difficulty.

Follow-Up Status

Among the 1,857 individuals with complete baseline data on all independent variables utilized in the analysis, 61.2% (n = 1,137) had outcome data for all four waves of data collection, 18.2% (n = 338) had outcome data for three of the four waves, 10.7% (n = 199) had outcome data for two waves, and 9.9% (n = 183) had outcome data for just one wave. All participants with data for at least one wave were included in the analysis. To permit investigation of the impact of incomplete follow-up on our findings, we created a dichotomous variable that indicated whether or not the respondent had incomplete follow-up across the four waves.

Statistical Analysis

To account for the fact that we have repeated measures of physical activity nested within individuals who are nested within sampling strata and sampling clusters, we used hierarchical linear modeling (HLM) to examine trajectories of physical activity over time. Specifically, we used a three-level logistic regression model to examine the effect of gender on the log odds of engaging in physical activity between 2004 and 2010.

Level 1 predictors included time (in years) and a time² term. Level 2 predictors included gender, baseline demographic and health characteristics, and follow-up status. Features of the sample design (i.e., clustering and stratification) were accounted for in Level 3. Specifically, as described by Heeringa, West, and Berglund (2010), we created a variable that captured each respondent’s sampling stratum and sampling cluster and utilized this new variable as a Level 3 identifier. To ensure that estimates reflect the population of older adults in the United States in 2004, the 2004 sampling weights provided by HRS were incorporated into the HLM algorithm.

To determine how best to model the relationship between physical activity and time, we first plotted the percentage of older adults engaging in physical activity by time. This plot suggested that the relationship between time and engagement in physical activity was a curvilinear one, with the percentage of adults reporting engagement in physical activity becoming increasingly smaller with time. Thus, our initial mixed model included time and time² as Level 1 predictors of the log odds of engaging in physical activity.

To determine if there was evidence of significant between-person and between-stratum-cluster variation in the intercept and the effects of time and time², we also included random coefficients for the intercept and slopes for time and time² in our initial model. Results revealed significant Level 2 (p = .016) and Level 3 (p < .001) random effects for the intercept term, but no significant Level 2 or Level 3 random effects for the time or time² slopes (p > .500 in all cases). For this reason, the Level 2 and Level 3 random effects for time and time² were dropped from the model.

After establishing how best to model time, we then examined the unadjusted relationship between gender and the log odds of engagement in physical activity. Specifically, we entered gender as a Level 2 predictor of the intercept and slopes for time and time². In a series of subsequent models, we examined the relationship between gender and physical activity after adjusting for demographic characteristics, health characteristics, and follow-up status.

Results

Sample Characteristics

Table 1 contains weighted characteristics for the sample in 2004. Mean age of the sample was 74.1 years (SE = .2); 53.3% of the sample was female, 8.8% was Hispanic, 12.4% was non-Hispanic Black, and 78.8% was non-Hispanic White. More than two thirds (68%) of the sample reported a high school or higher education. With respect to health status, nearly one fifth of the sample reported a clinically significant level of depressive symptoms, approximately 86% reported one or more chronic conditions in addition to their diabetes, and more than 70% reported one or more mobility limitations. Mean cognitive score was 20.9 (SE = .2), where a score of 10 or below is suggestive of cognitive impairment (Langa et al., 2008). At baseline, less than half of the sample (45.5%) reported engaging in moderate or vigorous physical activity more than once a week.

Table 1.

Sample Characteristics (n = 1,857).

Characteristic	Overall	Men	Women	p for gender differences
Characteristic	M (SE) or % (n)	M (SE) or % (n)	M (SE) or % (n)	p for gender differences
Age (years)	74.1 (.2)	73.9 (.3)	74.2 (.3)	.292
Gender
Men	46.7 (868)	—	—	—
Women	53.3 (989)
Education
Less than high school	32.1 (658)	27.6 (264)	35.9 (394)	<.0001
High school	51.2 (908)	50.0 (422)	52.2 (486)
Some college or higher	16.8 (291)	22.4 (182)	11.9 (109)
Race-ethnicity
Hispanic	8.8 (230)	7.3 (91)	10.1 (139)	<.001
Non-Hispanic Black	12.4 (361)	10.0 (141)	14.5 (220)
Non-Hispanic White	78.8 (1,266)	82.7 (636)	75.4 (630)
Wealth quartiles
Lowest	22.8 (463)	15.9 (141)	28.8 (322)	<.0001
Second lowest	24.0 (464)	21.6 (204)	26.1 (260)
Second highest	25.4 (465)	26.1 (239)	24.7 (226)
Highest	27.8 (465)	36.3 (284)	20.4 (181)
Cognitive score	20.9 (.2)	21.1 (.2)	20.8 (.2)	.271
Chronic conditions
None	13.8 (255)	16.5 (146)	11.4 (109)	.007
One	34.6 (660)	34.2 (308)	34.9 (352)
Two	32.5 (604)	29.7 (257)	35.0 (347)
Three or more	19.1 (338)	19.6 (157)	18.7 (181)
≥4 Depressive symptoms	19.0 (368)	13.0 (117)	24.3 (251)	<.0001
Mobility limitations
None	28.5 (528)	39.2 (333)	19.2 (195)	<.0001
One	21.2 (395)	19.6 (171)	22.6 (224)
Two	16.8 (313)	15.3 (135)	18.1 (178)
Three	11.5 (217)	10.3 (90)	12.5 (127)
Four	13.2 (245)	10.9 (92)	15.3 (153)
Five	8.7 (159)	4.7 (47)	12.3 (112)
Physically active	45.5 (813)	51.6 (433)	40.1 (380)	<.0001

Note. Percentages are weighted. Sample numbers are not weighted. Percentages may not sum to 100 due to rounding.

Whereas no significant gender differences were observed for age (p = .292) or cognitive score (p = .271), significant gender differences were evident for education, race and ethnicity, level of wealth, number of chronic conditions, depressive symptoms, and mobility limitations (p < .01 in all cases). Notable differences were also evident for baseline levels of physical activity, with approximately one half of men engaging in moderate or vigorous physical activity compared with about 40% of women (p < .0001).

Trajectories of Physical Activity

Table 2 contains the results of our HLM analysis. As shown in Model 1, in which we examined the effect of time and time² on the log odds of engaging in physical activity, the linear term for time was not significant (p = .840). Time², however, was statistically significant (p < .001) and negative, suggesting a decreasing probability of engaging in physical activity over time among older adults with diabetes. Based on the coefficients for Model 1, the predicted probability of engaging in physical activity declined by 34% between baseline and Year 6.

Table 2.

Multilevel Models Examining Physical Activity Over Time Among Older U.S. Adults With Diabetes.

Parameter	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Fixed effects
Intercept	−.178	−.444	−.470	.264	.478	.731
Intercept	(.085)*	(.115)***	(.098)***	(.164)	(.354)	(.354)*
Level 1
Time	.008	−.026	.013	.007	−.012	−.022
Time	(.042)	(.056)	(.042)	(.043)	(.045)	(.044)
Time²	−.020	−.014	−.020	−.021	−.020	−.021
Time²	(.006)***	(.008)	(.006)***	(.006)***	(.006)**	(.006)***
Level 2
Baseline age				−.036	−.017	−.009
Baseline age				(.008)***	(.009)	(.009)
Education (Referent = ≥ some college)
Less than high school				−.533	−.205	−.194
Less than high school				(.162)**	(.141)	(.14)
High school				−.141	−.023	−.024
High school				(.140)	(.122)	(.118)
Male (Referent = female)		.562	.617	.442	.205	.256
Male (Referent = female)		(.128)***	(.089)***	(.083)***	(.091)*	(.092)**
Race-ethnicity (Referent = White)
Black				.017	−.019	−.042
Black				(.109)	(.111)	(.114)
Hispanic				.047	.041	.018
Hispanic				(.127)	(.147)	(.145)
Wealth quartile (Referent = highest)
Lowest				−.844	−.364	−.352
Lowest				(.145)***	(.140)**	(.141)*
Second lowest				−.572	−.247	−.246
Second lowest				(.130)***	(.123)*	(.124)*
Second highest				−.314	−.143	−.135
Second highest				(.120)**	(.112)	(.111)
Chronic conditions (Referent = none)
One					−.083	−.083
One					(.144)	(.141)
Two					−.210	−.205
Two					(.151)	(.148)
Three or more					−.215	−.200
Three or more					(.191)	(.187)
≥4 Depressive symptoms (Referent = 3 or less)					−.214	−.215
≥4 Depressive symptoms (Referent = 3 or less)					(.115)	(.117)
Mobility limitations					−.476	−.457
Mobility limitations					(.027)***	(.028)***
Cognitive functioning					.024	.018
Cognitive functioning					(.011)*	(.011)
Incomplete follow-up						−.539
Incomplete follow-up						(.082)***
Interactions
Gender × Time		.081
Gender × Time		(.072)
Gender × Time²		−.013
Gender × Time²		(.012)
Random effects	Variance
Level 2, r_0ij	1.496***	1.425***	1.425***	1.30***	.996***	.984***
Level 3, u_00j	.054**	.056**	.056**	.056***	.031*	.028*

Note. Standard errors are in parentheses.

p < .05. **p < .01. ***p < .001.

In our next model, gender was entered as a Level 2 predictor of the intercept and slopes for time and time². Results revealed a significant main effect of gender (p < .001), but no significant gender by time (p = .256) or gender by time² (p = .270) interaction, suggesting that the effect of time on the log odds of engaging in physical activity did not vary by gender (see Model 2). Because there was no evidence of a significant gender by time interaction, gender was removed as a Level 2 predictor of the slopes for time and time² and the model was re-run. Exponentiating the coefficient for gender in the revised model (see Model 3 in Table 2) produced an odds ratio (OR) of 1.85 (95% CI = [1.56, 2.21]), indicating that the odds of men engaging in physical activity were 85% higher than for women (p < .001). In the absence of a significant gender by time interaction, this means that the physical activity trajectory for men was consistently more favorable than the trajectory for women, with no evidence of widening or narrowing of the gender gap over time (see Figure 1).

Figure 1.

Estimated probability of engagement in physical activity by gender: 2004-2010.

Although controlling for baseline demographic factors reduced the effect of gender on the log odds of engaging in physical activity (see Model 4 in Table 2), gender remained a significant predictor of physical activity. Exponentiating the coefficient for gender in Model 4 produced an OR of 1.56 (95% CI = [1.32, 1.83]), indicating that the odds of engaging in physical activity were 56% higher for men than women after adjusting for other demographic characteristics.

Controlling for cognitive score, depressive symptoms, mobility limitations, and number of comorbid health conditions further reduced the effect of gender (see Model 5). Nevertheless, gender remained a significant predictor of the log odds of engaging in physical activity. After adjusting for both baseline demographic and health characteristics, the odds of engaging in physical activity were 23% higher for men than for women (OR = exp (.205) = 1.23; 95% CI = [1.03, 1.47]).

Finally, we investigated the impact of incomplete follow-up on the relationship between gender and the log odds of engaging in physical activity. To do so, we entered follow-up status and a gender by follow-up interaction term into the model as Level 2 predictors of the intercept. Results indicated a significant main effect of follow-up status (p < .001) on the log odds of engaging in physical activity, but no significant gender by follow-up interaction (p = .336; model results not displayed). We, therefore, dropped the gender by follow-up interaction term from the model. As shown in Model 6 in Table 2, the results of our final model indicated that the odds of men engaging in physical activity remained significantly higher than for women after controlling for follow-up status. Specifically, the odds of engaging in physical activity were 29% higher for men than for women (OR = exp (0.256) = 1.29; 95% CI = [1.08,1.55]).

Discussion

In this investigation, we examined engagement in physical activity over a 6-year period among older U.S. adults with diabetes and the impact of gender on physical activity over time. We found that less than half of older adults with diabetes were physically active at baseline, with the percentage of men engaging in physical activity greater than for women. Over the 6-year period, the probability of engaging in physical activity declined by more than 30%, with no significant variation by gender. Although men and women experienced similar declines in physical activity, the physical activity trajectory for women started and remained less favorable than for men.

Our finding that older women were less likely to engage in physical activity than older men is consistent with published findings for those with and without diabetes (Zhao et al., 2011). In our sample of older adults, the odds of men engaging in physical activity were 85% higher than for women. Not surprisingly, the size of the observed gender difference diminished after controlling for demographic and health covariates known to be associated with both gender and physical activity. Importantly, however, a significant gender gap remained after controlling for these factors, with the odds of women engaging in physical activity approximately one-quarter lower than for men. This finding is similar to that reported by Zhao et al. (2011) in their examination of data from the Behavioral Risk Factor Surveillance Survey. After controlling for a range of potential covariates, they found that the odds of meeting physical activity guidelines were 20% to 24% lower for older women with diabetes relative to their male counterparts, depending on the guidelines used.

Gender-related norms for physical activity may be one reason for the gender gap observed in the current study. As noted previously, in earlier periods of life, many older women were exposed to normative beliefs that discouraged women and girls from engaging in vigorous physical activity (Lutter, 1994). Given that engagement in physical activity in earlier phases of life is associated with physical activity in older adulthood (Hirvensalo & Lintunen, 2011), part of the observed gender gap may be the end result of being socialized to limit engagement in vigorous activity. Data from the National Health Interview Survey (NHIS) suggest, however, that the gender patterns observed in this cohort are not merely a legacy of being raised in an era with more rigidly defined gender norms. More specifically, 2008-2010 NHIS data reveal gender differences in leisure-time physical activity across adult age categories, with the percentage of males engaged in “high activity” greater than females even among the youngest adults (i.e., those aged 18-24; Schoenborn, Adams, & Peregoy, 2013).

Although men are generally reported to be more physically active than women, it is important to recognize that engagement in physical activity is not static across the life course. Research suggests that life events, transitions, and social roles can influence engagement in physical activity, with some events, transitions, and roles (e.g., being a caregiver; Connell, 1994; Schulz et al., 1997) making it more difficult to engage in physical activity and others (e.g., retirement) making it easier to do so (Barnett, van Sluijs, & Ogilvie, 2012; Brown et al., 2009; Hirvensalo & Lintunen, 2011). Given that role expectations often vary by gender (Bird & Rieker, 2008), transitioning into and out of various roles may result in a widening or narrowing of the gender gap in physical activity at various points over the life course. Although the data are cross-sectional and need to be interpreted with caution, 2003-2004 accelerometry counts from the National Health and Nutrition Examination Survey suggest that the gender gap in physical activity is wider at some ages (e.g., 30-39) than others (e.g., 60-69; Troiano et al., 2008). Future research that captures a larger portion of the life course would enhance our understanding of how the gender gap in physical activity changes over the life course and shed insight into the events and transitions that may contribute to differences in physical activity among older men and women with diabetes.

Although we controlled for the presence of commonly reported health problems, it should be noted that at least some of the remaining gender gap may be the result of residual confounding by health. We were unable to control for the severity of functional limitations, for example, due to limitations of the existing measures. If the severity of functional limitations varies by gender, merely controlling for the number of limitations will not fully account for gender differences in functional health problems. Likewise, although we controlled for number of chronic conditions, we did not examine the effect of specific conditions on physical activity. Existing evidence suggests that the presence of conditions such as arthritis (CDC, 2008) and heart disease (Janevic et al., 2013; Zhao et al., 2011), which vary by gender, exert negative effects on physical activity in the context of diabetes. Future research on gender differences would benefit from a more nuanced examination of health problems and their cumulative and complex impact on health behavior.

The observed decline in physical activity over time for both men and women is consistent with recent research on the general population of older U.S. adults. Using data from the Americans’ Changing Lives study, Shaw and colleagues (2010) observed a steady decline in engagement in physical activity among older adults over a 16-year period. In contrast with our investigation, they found that the declines were steeper for older women than for older men—a finding that was not statistically significant once they accounted for gender differences in health.

What might explain the observed decline in physical activity over time among men and women with diabetes? Although it was not examined in this investigation, the onset of diabetes-related complications could have contributed to the decline. Using data from a supplement to the HRS, Janevic et al. (2013) found retinopathy, nephropathy, and neuropathy to be independently associated with a reduced odds of meeting physical activity guidelines among adults aged 51 years and over with diabetes. The onset of such complications may create additional barriers to physical activity, making it increasingly less likely that an older person will engage in physical activity across time. Likewise, over 80% of the older adults in this sample reported a chronic condition in addition to their diabetes. Part of the reason for the observed decline in physical activity may be that it becomes more difficult to engage in regular physical activity when living with multiple health conditions. Qualitative work suggests that some adults with multimorbidity find that the symptoms or management needs of one condition take precedence over the management needs of another condition (Bayliss, Steiner, Fernald, Crane, & Main, 2003; Morris, Sanders, Kennedy, & Rogers, 2011), suggesting less than optimal management of all conditions.

Although this study adds to our understanding of physical activity among older adults with diabetes, there are a number of limitations in this investigation that need to be taken into account. First, the items used to assess physical activity in the HRS are less than ideal. In addition to being self-report measures that may not reflect actual levels of physical activity, the items do not contain information about duration of physical activity and only crudely capture frequency of engagement. It is, therefore, likely that a number of older adults classified as being physically active did not actually meet the ADA guidelines. Although our estimate of the percentage of physically active adults (45.5%) is similar to what Janevic et al. (2013) reported (43.1%) using a more detailed measure of physical activity administered as part of the 2003 HRS Mail Survey on Diabetes, it is considerably higher than Zhao et al.’s estimate of the percentage meeting the 2007 ADA guidelines for physical activity (25%). In the latter case, adults were required to report engaging in physical activity 3 or more days each week. It should also be acknowledged that the error associated with these measures may vary by gender. To the extent that this is true, the relationship between gender and physical activity observed in this study may be biased. Our finding that women are less likely to be engaged in physical activity than men, however, is consistent with prior research on physical activity in older adults with diabetes that involved a different measure of physical activity (Zhao et al., 2011). Second, this investigation included only older adults and does not capture earlier phases of the life course. To more fully understand the influence of gender on physical activity, studies that follow individuals over a longer period of the life course are needed. Third, transitions in health such as the onset of diabetes complications were not examined in this study. Given that existing research indicates that diabetes complications are associated with a lower likelihood of engaging in physical activity, future research should be directed at understanding how these and non-health transitions (e.g., the transition to retirement) influence physical activity in the context of diabetes. Despite these limitations, we believe that this study is an important first step in exploring the impact of gender on physical activity among older adults with diabetes. Notable strengths of this study include the large national sample, the use of multiple waves of physical activity data covering a 6-year period, and the availability of an array of measures that allowed us to partially explain the observed gender differences in physical activity.

By 2050, as many as one in three U.S. adults are projected to have diabetes (Boyle, Thompson, Gregg, Barker, & Williamson, 2010). Although this is of concern, proper management of the condition can prevent or delay some of the complications of the disease. Physical activity is key to diabetes management. Unfortunately, a sizable percentage of older adults with diabetes do not engage in regular physical activity and the probability of doing so declines with time. Although the percentage of physically active adults is suboptimal for both men and women, the older women in this cohort started and remained less active than men over time. Our understanding of how best to promote physical activity among both older men and women with diabetes would be enhanced by longitudinal research covering larger portions of the life course and investigations that identify life events and transitions that may serve as vantage points for intervention.

Footnotes

Authors’ Note

The following data files were utilized in this analysis: 2004 Core Final, Version 1.0; 2006 Core Final, Version 2.0; 2008 Core Final, Version 2.0; 2010 Core Final, Version 3.0; HRS Tracker 2010 File; RAND HRS Data (Version M); COGIMP9210A_R.da (cognitive data).

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: This research was funded through a grant from the University of Michigan’s Sport, Health, and Activity Research and Policy (SHARP) Center. The Health and Retirement Study is sponsored by the National Institute on Aging (Grant number NIA U01AG009740) and is conducted by the University of Michigan.

References

Barnett

van Sluijs

E. M.

Ogilvie

(2012). Physical activity and transitioning to retirement: A systematic review. American Journal of Preventive Medicine, 43, 329-336. doi:10.1016/j.amepre.2012.05.026

Bayliss

E. A.

Steiner

J. F.

Fernald

D. H.

Crane

L. A.

Main

D. S.

(2003). Descriptions of barriers to self-care by persons with comorbid chronic diseases. Annals of Family Medicine, 1, 15-21. doi:10.1370/afm.4

Bird

C. E.

Rieker

P. P.

(2008). Gender and health: The effects of constrained choices and social policies. Cambridge, UK: Cambridge University Press.

Boyle

J. P.

Thompson

T. J.

Gregg

E. W.

Barker

L. E.

Williamson

D. F.

(2010). Projection of the year 2050 burden of diabetes in the US adult population: Dynamic modeling of incidence, mortality, and prediabetes prevalence. Population Health Metrics, 8, 29. doi:10.1186/1478-7954-8-29

Brown

W. J.

Heesch

K. C.

Miller

Y. D.

(2009). Life events and changing physical activity patterns in women at different life stages. Annals of Behavioral Medicine, 37, 294-305. doi:10.1007/s12160-009-9099-2

Centers for Disease Control and Prevention. (2008). Arthritis as a potential barrier to physical activity among adults with diabetes—United States, 2005 and 2007. Morbidity and Mortality Weekly Report Surveillance Summaries, 57, 486-489.

Centers for Disease Control and Prevention. (2014). National diabetes statistics report: Estimates of diabetes and its burden in the United States, 2014. Atlanta, GA: U.S. Department of Health and Human Services. Retrieved from http://www.cdc.gov/diabetes/pubs/statsreport14/national-diabetes-report-web.pdf

Chiu

C. J.

Wray

L. A.

(2011a). Gender differences in functional limitations in adults living with type 2 diabetes: Biobehavioral and psychosocial mediators. Annals of Behavioral Medicine, 41, 71-82. doi:10.1007/s12160-010-9226-0

Chiu

C. J.

Wray

L. A.

(2011b). Physical disability trajectories in older Americans with and without diabetes: The role of age, gender, race or ethnicity, and education. The Gerontologist, 51, 51-63. doi:10.1093/geront/gnq069

10.

Colberg

S. R.

Sigal

R. J.

Fernhall

Regensteiner

J. G.

Blissmer

B. J.

Rubin

R. R.

, . . . American Diabetes Association. (2010). Exercise and type 2 diabetes: The American College of Sports Medicine and the American Diabetes Association: Joint position statement. Diabetes Care, 33(12), e147-e167. doi:10.2337/dc10-9990

11.

Conn

V. S.

(1998). Older women’s beliefs about physical activity. Public Health Nursing, 15, 370-378. doi:10.1111/j.1525-1446.1998.tb00362.x

12.

Connell

C. M.

(1994). Impact of spouse caregiving on health behaviors and physical and mental health status. American Journal of Alzheimer’s Disease & Other Dementias, 9, 26-36. doi:10.1177/153331759400900105

13.

Crimmins

E. M.

Kim

J. K.

Solé-Auró

(2011). Gender differences in health: Results from SHARE, ELSA and HRS. European Journal of Public Health, 21, 81-91. doi:10.1093/eurpub/ckq022

14.

Engelgau

M. M.

Geiss

L. S.

Saaddine

J. B.

Boyle

J. P.

Benjamin

S. M.

Gregg

E. W.

. . .Narayan

K. M.

(2004). The evolving diabetes burden in the United States. Annals of Internal Medicine, 140, 945-950. doi:10.7326/0003-4819-140-11-200406010-00035

15.

Exalto

L. G.

Whitmer

R. A.

Kappele

L. J.

Biessels

G. J.

(2012). An update on type 2 diabetes, vascular dementia and Alzheimer’s disease. Experimental Gerontology, 47, 858-864. doi:10.1016/j.exger.2012.07.014

16.

Federal Interagency Forum on Aging-Related Statistics. (2012). Older Americans 2012: Key indicators of well-being. Washington, DC: U.S. Government Printing Office.

17.

Fowler

M. J.

(2008). Microvascular and macrovascular complications of diabetes. Clinical Diabetes, 26, 77-82. doi:10.2337/diaclin.26.2.77

18.

Heeringa

S. G.

Connor

J. H.

(1995). Technical description of the Health and Retirement Survey sample design. Retrieved from http://hrsonline.isr.umich.edu/sitedocs/userg/HRSSAMP.pdf

19.

Heeringa

S. G.

West

B. T.

Berglund

P. A.

(2010). Advanced topics in the analysis of survey data. Applied survey data analysis (pp. 371-398). Boca Raton, FL: Chapman & Hall/CRC.

20.

Herzog

A. R.

Wallace

R. B.

(1997). Measures of cognitive functioning in the AHEAD Study. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 52(Spec. No.), 37-48.

21.

Hirvensalo

Lintunen

(2011). Life-course perspective for physical activity and sports participation. European Review of Aging and Physical Activity, 8, 13-22. doi:10.1007/s11556-010-0076-3

22.

Janevic

M. R.

McLaughlin

S. J.

Connell

C. M.

(2013). The association of diabetes complications with physical activity in a representative sample of older adults in the United States. Chronic Illness, 9, 251-257. doi:10.1177/1742395313475461

23.

Juster

F. T.

Suzman

(1995). An overview of the Health and Retirement Study. Journal of Human Resources, 30, S7-S56. doi:10.2307/146277

24.

Laditka

S. B.

Laditka

J. N.

(2015). Active life expectancy of Americans with diabetes: Risks of heart disease, obesity, and inactivity. Diabetes Research and Clinical Practice, 107, 37-45. doi:10.1016/j.diabres.2014.10.008

25.

Langa

K. M.

Larson

E. B.

Karlawish

J. H.

Cutler

D. M.

Kabeto

M. U.

Kim

S. Y.

Rosen

A. B.

(2008). Trends in the prevalence and mortality of cognitive impairment in the United States: Is there evidence of a compression of cognitive morbidity? Alzheimer’s & Dementia, 4, 134-144. doi:10.1016/j.jalz.2008.01.001

26.

Lutter

J. M.

(1994). History of women in sports. Societal issues. Clinical Journal of Sport Medicine, 13, 263-279.

27.

Morris

R. L.

Sanders

Kennedy

A. P.

Rogers

(2011). Shifting priorities in multimorbidity: A longitudinal qualitative study of patient’s prioritization of multiple conditions. Chronic Illness, 7, 147-161. doi:10.1177/1742395310393365

28.

Nothwehr

Stump

(2000). Health-promoting behaviors among adults with type 2 diabetes: Findings from the Health and Retirement Study. Preventive Medicine, 30, 407-414. doi:10.1006/pmed.2000.0658

29.

Ofstedal

M. B.

Fisher

G. G.

Herzog

A. R.

(2005). Documentation of cognitive functioning measures in the Health and Retirement Study. Retrieved from http://hrsonline.isr.umich.edu/sitedocs/userg/dr-006.pdf

30.

Palmer

R. F.

Espino

D. V.

Dergance

J. M.

Becho

Markides

(2012). The role of physical activity and diabetes status as a moderator: Functional disability among older Mexican Americans. Age Ageing, 41, 752-758. doi:10.1093/ageing/afs106

31.

Roy

Lloyd

C. E.

(2012). Epidemiology of depression and diabetes: A systematic review. Journal of Affective Disorders, 142(Suppl.), S8-S21. doi:10.1016/S0165-0327(12)70004-6

32.

Schoenborn

C. A.

Adams

P. F.

Peregoy

J. A.

(2013). Health behaviors of adults: United States, 2008-2010. Vital and Health Statistics, 10(257), 1-184.

33.

Schulz

Newsom

Mittelmark

Burton

Hirsch

Jackson

(1997). Health effects of caregiving: The caregiver health effects study: An ancillary study of the Cardiovascular Health Study. Annals of Behavioral Medicine, 19, 110-116. doi:10.1007/BF02883327

34.

Shaw

B. A.

Liang

Krause

Gallant

McGeever

(2010). Age differences and social stratification in the long-term trajectories of leisure-time physical activity. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 65, 756-766. doi:10.1093/geronb/gbq073

35.

Steffick

D. E.

(2000). Documentation of affective functioning measures in the Health and Retirement Study. Retrieved from http://hrsonline.isr.umich.edu/sitedocs/userg/dr-005.pdf

36.

Stessman

Jacobs

J. M.

(2014). Diabetes mellitus, physical activity, and longevity between the ages of 70 and 90. Journal of the American Geriatrics Society, 62, 1329-1334. doi:10.1111/jgs.12930

37.

Troiano

R. P.

Berrigan

Dodd

K. W.

Mâsse

L. C.

Tilert

McDowell

(2008). Physical activity in the United States measured by accelerometer. Medicine & Science in Sports & Exercise, 40, 181-188. doi:10.1249/mss.0b013e31815a51b3

38.

Vrazel

Saunders

R. P.

Wilcox

(2008). An overview and proposed framework of social-environmental influences on the physical-activity behavior of women. American Journal of Health Promotion, 23, 2-12. doi:10.4278/ajhp.06070999

39.

Wilcox

Oberrecht

Bopp

Kammermann

S. K.

McElmurray

C. T.

(2005). A qualitative study of exercise in older African American and white women in rural South Carolina: Perceptions, barriers, and motivations. Journal of Women & Aging, 17, 37-53. doi:10.1300/J074v17n01_04

40.

Zhao

Ford

E. S.

Balluz

L. S.

(2011). Physical activity in U.S. older adults with diabetes mellitus: Prevalence and correlates of meeting physical activity recommendations. Journal of the American Geriatrics Society, 59, 132-137. doi:10.1111/j.1532-5415.2010.03236.x