Type 2 diabetes is a major cause of illness and death in the United States, especially among older adults, resulting in human and economic costs (1). In 2010, a total of 10.9 million Americans aged 65 years or older had either diagnosed or undiagnosed diabetes (1). Among these older adults, approximately 390,000 new diabetes cases were diagnosed in 2010 (1), and diagnosed diabetes cases will likely increase to 26.7 million in 2050 from 6.3 million in 2005 (2).

Adults with diabetes are at 1.8 to 2.4 times greater risk of various forms of disability than their nondiabetic counterparts (3). Older adults with diabetes are more likely than those without the disease to have less muscle strength, poorer muscle quality, and accelerated muscle loss (4). Older diabetic adults may respond favorably to exercise programs that improve glycemic control (5,6). Among older diabetic adults, resistance training can improve muscle strength and glycemic control (7,8). In addition, resistance training can prevent or delay the development of sarcopenia and osteoporosis and improve functional capacity in tasks of daily living among older adults with diabetes (6).

Older diabetic adults are more likely to have a sedentary lifestyle (9) and are less likely to comply with physical activity recommendations (10). A recently developed Reference Guide of Physical Activity for Older Adults (PARG) identifies and promotes physical activity programs for older adults (11) whose health status ranges from homebound to healthy.

Our objective was to provide state-specific estimates of inactive, older diabetic adults, categorized by defined health status groups. Knowing state-specific proportions in these groups may help planners select physical activity programs for their states (11).

Using cross-sectional data from the 2000 through 2009 National Health Interview Survey (NHIS), we identified health status categories to correspond to the health status groups defined by the PARG (11) (Box). Health status data for these categories are unavailable in the 2000 through 2009 Behavioral Risk Factor Surveillance System (BRFSS); therefore, we applied the NHIS regional estimates of the health status categories among inactive, older, diabetic adults to state-specific BRFSS estimates of older diabetic adults to create state estimates for each health status group.

BRFSS percentage distributions by age and sex resembled those of NHIS for similar years but differed by education and race (Table). In the 2000 through 2009 NHIS, the overall percentage of inactive older diabetic adults who were homebound was approximately 3% (95% CI, 3.4% (95% CI, 2.9%-3.9%); frail, 9.6% (95% CI, 8.9%-10.3%); impaired, 29.8% (95% CI, 28.7%-30.9%); having a comorbid condition, 50.5% (95% CI, 49.4%-51.8%); and healthy, 6.7% (95% CI, 6.1%-7.3%).

The NHIS regional percentages in the health status groups varied significantly for the frail, people with comorbid conditions, and the healthy (Figure 1). Specifically, the percentages for the frail status ranged from 8% in the Northeast to 11% in the South. Forty-nine percent in the South had at least 1 comorbid condition, compared with 55% in the Northeast. The percentage of the healthy ranged from 5% in the Midwest to 9% in the West.

None of the state-specific percentages of the homebound differed significantly among the states (range, 2.2%-3.0%) (Figure 2a). State-specific percentages by health status group varied significantly for the frail (6% in Massachusetts to 9% in Florida) (Figure 2b); the impaired (20% in Alaska to 26% in Iowa) (Figure 2c); people with at least 1 comorbid condition (35% in Texas to 44% in Pennsylvania) (Figure 2d); and people who are otherwise healthy (4% in Michigan and Ohio to 7% in Montana) (Figure 2e).

The estimated total of adults in each health status group tended to be lowest in the least populous state, Alaska, and highest in the most populous state, California (Appendices A and B). For Alaska, estimates ranged from 396 homebound people to 6,157 people having comorbid conditions (Figures 2a-2e). In California, estimates ranged from 35,598 homebound people to 552,807 people having comorbid conditions.

This study provides the first set of state-specific estimates of the percentages and the number of inactive, older, diabetic adults in 5 health status groups who may benefit from physical activity programs. These estimates may facilitate planning of physical activity programs for diabetic older adults with varying functional health status (11).

The highest percentage of the homebound and the frail corresponds with higher diagnosed diabetes rates in Southern states (16), and the highest percentage of the healthy corresponds with lower diagnosed diabetes rates in Western states (16). Our estimates may prove valuable to state planners because the functional burden among older adults with diabetes is greater than that for those without diabetes (17). People with diabetes in the United States are 2 to 3 times more likely than older, nondiabetic adults to have various forms of disability, including reduced lower-extremity mobility and reduced general physical activities (3). Followed prospectively, older diabetic adults have a 42% higher risk for developing disability than older nondiabetic adults (18). Diabetes often increases the risk for many other comorbid conditions (19) that may lead to future declines in physical function (18,20-22) and poorer health-related quality of life (23).

Physical activity programs can help older diabetic adults by improving glycemic control (24,25), which reduces the development of microvascular and other complications (5,25) and other comorbid conditions such as high blood pressure and dyslipidemia (5,25). Appropriate physical activity programs can restore function among homebound, frail, and impaired adults; control comorbid conditions; and help adults who have diabetes but are otherwise healthy to resist progression to a more severe health status. Recent guidelines for older adults call for individualized patient care and education, the prevention and control of cardiovascular risk factors, and screening for combinations of comorbid conditions, while also regulating glycemic status to prevent and control microvascular complications (26). These guidelines may require balancing glycemic targets, life expectancy, functional status, and patient preferences. Recognizing the proportion and number of older diabetic adults in each of the 5 health status groups is the first step toward devising strategies to prevent worsening and perhaps to improve their health status.

Our state-specific synthetic estimates for the 5 health status groups are, to our knowledge, the first to correspond with the interventions in the PARG (11). Planners may also choose from other physical activity programs to balance the proportion and the total number of adults in each health status group against implementation needs and costs. For example, programs for the few homebound or frail adults will likely require more expensive trained staff to focus on improving and preserving physical function and range of motion; the homebound will likely require a program delivered in their homes, whereas the frail often need transportation to a facility where a program is held. Impaired adults may be able to use their own transportation to get to a community program facility. Adults who have comorbid conditions may require monitoring to avoid complications from physical activity programs. For example, older diabetic adults who use beta blockers to control hypertension may be less attentive to the symptoms of hypoglycemia than those not taking such medications, and their muted exercise heart rate response could lead to overexertion if they attempt to meet a traditional target heart rate applicable to healthy older adults (27). Diabetic adults who are otherwise healthy should be encouraged to be active to improve glycemic control, to avoid hypoglycemia, and to maintain physical function (5,24). Program planners should consider these and many other factors identified in the PARG to deliver safe and successful programs (11) for specific health status groups of inactive, older diabetic adults.

Our study had several limitations. We used the statistically conservative method of nonoverlapping confidence intervals to determine significance and assumed, within region, that states did not vary in their proportions of inactive, older, diabetic adults by specific health status group. This conservative approach has the advantage of partially controlling for multiple comparisons. We used self-report items of impairment and disability from NHIS rather than a detailed geriatric assessment protocol (28) that might better ascertain each health status. Because we used self-report for both health status groups and physical inactivity, the study's exposure (physical inactivity) and outcome (health status groups) are subject to misclassification. The outcome was not available in representative data sets. We also operationally defined the 5 health status groups primarily to correspond to terms found in the PARG (11). Some of the terms, such as "frail," are hard to define conceptually or operationally and are subject to various interpretations (29). Moreover, we did not replicate the multiple categories of disability recently outlined by Kalyani and colleagues (3), who also used NHIS data. They analyzed all people with diabetes, regardless of activity, which ruled out a comparison with our analyses of people with diabetes who were inactive. Another limitation is that we did not identify separate health status groups for the roughly 1 in 4 physically active, older diabetic adults. However, their need for physical activity programs is not as pressing as for the remaining inactive, older diabetic adults because the combined proportion of the homebound and the frail was higher for the inactive, 12.9% (homebound + frail = 3.3% + 9.6%, respectively), than for the active, 1.8% (homebound + frail = 0.5% + 1.3%, respectively). Nonetheless, active older adults with diabetes may benefit from a formal physical activity program. Lastly, we did not adjust our estimates for age or any other variable that might have produced some of the regional variation in our estimates because our goal was to present data for health status group by state. Age and other factors that differ among states might change the size of our estimates. However, because we stratified our analyses and presented data for only people who were physically inactive, we did reduce regional variation in health status groups.

In summary, among older, physically inactive, diabetic adults, the homebound, the frail, and the otherwise healthy were less common than those who were impaired or who had comorbid conditions. Tailored physical activity programs have much to offer older, physically inactive, diabetic adults. State program planners may use these estimates to select physical activity programs for each kind of health status group to enhance diabetes control and to prevent adverse health outcomes that lead to progressively worse impairment. Making such programs available may help to control the preventable burden of diabetes that is otherwise expected to grow unabated with the anticipated expansion of the older adult diabetic population. Because of this expansion, we anticipate that these state-level numbers may also change; therefore, future studies may investigate changes in these estimates as well as urban-rural and racial differences that may exist in health status groups among the states.