Aging and Physical Function in Type 2 Diabetes: 8 Years of an Intensive Lifestyle Intervention

Abstract

Background.

Compared with adults without type 2 diabetes mellitus, those with the disease experience more limitations in their physical functioning (PF). Look AHEAD is a large multicenter trial that examined the effects of an intensive lifestyle intervention (ILI) for weight loss on cardiovascular outcomes compared with diabetes support and education (DSE). Although the current study compared treatment differences between ILI and DSE on PF, the primary goal was to examine whether this effect was moderated by age and history of cardiovascular disease at enrollment.

Methods.

Overweight or obese adults with type 2 diabetes mellitus (n = 5,145) were randomly assigned to either ILI or DSE. The mean (±SD) age and % females in ILI was 58.9 years (±6.9) and 59.8%; it was 58.6 years (6.8) and 59.5% in DSE. Analysis in 4,998 participants assessed the differential rates of decline in PF across a period of 8 years for the ILI and DSE groups.

Results.

ILI resulted in improved PF compared with DSE after 1 year (p < .0001) and was maintained across time. Within the ILI, older adults experienced greater improvements than younger adults (p < .0001). By year 2, persons in ILI with preexisting cardiovascular disease were no different in PF than in DSE participants with preexisting cardiovascular disease.

Conclusion.

With the exception of persons who had a history of cardiovascular disease, ILI slowed the decline in PF with type 2 diabetes mellitus despite weight regain, an effect that was stronger for older than younger participants and could translate into reductions in falls and disability.

Type 2 diabetes mellitus (T2DM) is a serious threat to public health both in the United States (1,2) and globally (3). T2DM is largely a disease of obesity and aging (4), and its prevalence is expected to increase in the United States over the next two decades as a result of our overweight and aging population (5,6). The combined effect of T2DM and aging on physical functioning (PF) threatens independence of older adults (7), compromises glucose storage and clearance via loss of muscle mass (8), and reduces quality of life (9). Existing data reinforce the public health significance of this problem, in that older adults with T2DM have twice the prevalence of physical disability as those without the disease (10), and the severity of obesity further increases this risk (11). In the Look AHEAD study, a large multicenter trial examining the effects of an intensive lifestyle intervention (ILI) on cardiovascular events in T2DM, the first 4 years of the ILI treatment led to a 48% reduction in limitations with specific mobility-related activities compared with diabetes support and education (DSE) (12). In addition, across the first 8 years of the study, participants in the ILI treatment had higher scores on the SF-36 composite index for physical health than those in DSE (13).

The current study is a secondary data analysis of 8 years of data within Look AHEAD that examines self-reported PF using the 10-item subscale from the SF-36 composite physical health index. Although we first examine treatment differences between ILI and DSE, and expected to find that ILI would have higher PF than DSE, our primary goal was to examine whether age or history of cardiovascular disease (CVD) at enrollment moderated these treatment effects. Published results from both the Diabetes Prevention Program (14) and Look AHEAD (15) would suggest that older participants should derive more benefit in PF from the ILI treatment than younger participants. Also, based on the published results for the primary outcome paper in Look AHEAD (16), we anticipated that preexisting CVD would mitigate the benefit from ILI compared with those without preexisting CVD and that the decline in PF across time would be greater for participants with than participants without CVD.

Methods

Participants

As described in a previous methods paper (17), the Look AHEAD cohort included overweight or obese persons aged 45–76 at enrollment. Major exclusions included HbA1c > 11%, blood pressure > 160/100 mmHg, triglycerides > 600mg/dL, inadequate control of comorbid conditions, underlying diseases likely to limit life span or affect safety, and failure to pass a baseline graded maximal exercise stress test. At baseline, the cohort had limitations in mobility as determined by self-report (18) and performance on a treadmill test (19). Informed consent was obtained from all participants before screening. More details on the inclusion/exclusion criteria have been published (17), as has a CONSORT flow diagram for the first 8 years of the trial (16) (see Supplementary Appendix, Figure 1).

Figure 1.

Plot of % weight loss (Panel A) and physical activity in kcal/wk (Panel B) from baseline to follow-up visits at year 1, 4, and 8 for intensive lifestyle intervention (ILI) and diabetes support and education (DSE); for all time points in both PANELS, ILI differs from DSE at p < .0001.

Look AHEAD

Look AHEAD was a multicenter randomized controlled trial designed to determine whether intentional weight loss reduces cardiovascular morbidity and mortality in overweight/obese persons with T2DM. Participants were randomly assigned to ILI or DSE between October 2001 and ending in May 2004. The two primary goals for the ILI were to achieve a mean loss > 7% of initial weight and to increase physical activity to > 175min/wk. There were two phases of intervention during the first 4 years of follow-up. Months 1–6 of the first phase consisted of 24 sessions: three weekly group sessions followed by one individual session with a lifestyle counselor. For diet, participants were encouraged to replace breakfast and lunch with an energy drink for a target of 1,200–1,500 kcal/day for those 114 kg or less at baseline and a target of 1,500–1,800 kcal/day for those > 114 kg. Intervention relied heavily on unsupervised exercise at home and for most participants this consisted of brisk walking. Months 7–12 of the first phase consisted of two group sessions and one individual session. The second phase took place during years 2–4. This phase consisted of a minimum of two contacts/month with one being on-site and the other by phone, mail, or e-mail. More details on the key components of the ILI have been described (20).

DSE participants were offered three group sessions a year focusing on nutrition, physical activity, or social support, and the details of this comparison group have been described as well (21).

Assessments

Physical functioning.

To examine change in PF, we used the 10-item PF scale from the Medical Outcomes Survey Short-Form (SF-36) (22). This measure was completed at each of the eight annual assessment visits and is one of the most commonly used and well-validated measures in the epidemiological literature (23).

Leisure-time physical activity.

Leisure-time physical activity was assessed at baseline, Year 1, Year 4, and Year 8 on a subsample of participants in Look AHEAD using the Paffenbarger questionnaire (24); it was completed as part of a structured interview. The sample was based on selected clinical sites that included this questionnaire as part of their assessments. Data collected on the flights of stairs climbed, number of city blocks walked, and other fitness, sport, and recreational activities performed during the week prior to the assessment visit were used to compute kcal/wk of leisure-time physical activity.

Statistical Methods

Descriptive statistics for PF and other related variables were first calculated. The PF scale of the SF-36 is scored on a standardized T-scale, with a mean of 50 and a standard deviation of 10. Bivariate analysis included quartiles for the measure of PF by hypertension, CVD status, and other risk variables. To contrast the effects of ILI on patterns of change in PF with those in DSE, we used both exploratory analyses, especially graphs and visualization, and statistical tests of hypotheses. We used generalized estimating equation for model parameter estimation (25). The linear model can be expressed as follows:

$$\begin{array}{l}E(Y_{ik})={\beta }_0+{\beta }_1{\text{Interv}}_i+{\beta }_2{\text{Time}}_i\\ +{\beta }_3{\text{Interv}}_i\times {\text{Time}}_i+{\beta }_4{\text{BL}}_i,\end{array}$$

where $Y_{ik}$ , the measure of PF for individual i at the kth occasion; Interv, intervention status of either ILI or DSE; BL, measure of PF at baseline; and β, regression coefficient. Furthermore, in our preliminary analyses we found an abrupt increase in between-group differences that tended to disrupt the linearity assumption for change in PF over time as specified in the linear model. As a result, we modified the model to include an additional dummy variable for the first time point in the intervention group. Because Look AHEAD is a randomized controlled trial with a relatively large sample size, we did not include other baseline covariates in the model; the effect of the intervention on trends in PF across the 8 years was assessed through significance testing of the interaction term $\text{Interv}\times \text{Time}$ .

To examine the effect of two potential moderating variables (M)—age and CVD history—on PF, our analysis was based on the linear model specified previously with the added terms M, $M\times \text{Interv}$ , $M\times \text{Time}$ $M\times \text{Time}$ , and $M\times \text{Interv}\times \text{Time}$ . The effect of moderation was assessed through significance testing of the interaction term $M\times \text{Interv}$ , whereas the differential rates of decline between the older (CVD) and the younger (no CVD) intervention groups were assessed through significance testing of the term $M\times \text{Time}$ . The possible additional effect due to the interaction between age (CVD) and intervention on the rate of decline was assessed through testing of $M\times \text{Interv}\times \text{Time}$ .

Results

This study included 4,998 of 5,145 randomized participants (97%) who had data on the PF scale from at least 1 follow-up visit. The baseline characteristics of the 4,998 participants in the current study were very similar to those of the entire study cohort (Table 1) (26).

Table 1.

Baseline Demographic Characteristics: Mean (±SD) or Percentage

Variable	Diabetes Support and Education	Intensive Lifestyle Intervention
Number	2,490	2,508
Age	58.9±6.9	58.6±6.8
Sex (% F)	59.8%	59.5%
Race/Ethnicity
Black	15.6%	15.6%
American Indian Alaskan	5.0%	5.1%
Asian/Pacific islander	0.8%	1.2%
White	63.4%	63.1%
Hispanic	13.2%	13.1%
Other	2.0%	2.0%
Weight (kg)	100.8±18.9	100.6±19.7
Height (cm)	167.2±9.9	167.2±9.6
BMI	36.0±5.8	35.9±6.0
A1c (%)	7.30±1.19	7.25±1.15
Fitness (MET level)	7.18±1.98	7.20±1.95
SF-36 physical function	48.44 (7.97)	48.48 (7.83)
History of cardiovascular disease	13.5%	14.3%
Hypertension (%)	82.9%	84.1%
Smoking
Never	51.4%	49.7%
Past	44.4%	45.8%
Present	4.2%	4.6%
Oral hypoglycemic medication use	67.8%	68.0%
Insulin use	19.1%	18.5%
No diabetic medications	13.1%	13.5%
Knee pain	50.0%	50.0%

Notes: BMI = body mass index measured as weight in kg divided by height in m². Fitness is the estimated MET value from a graded exercise test.

Average weight loss across the 8-year period was greater in ILI (n = 2,160) than in DSE (n =2,206): 5.29% versus 2.70%; p < .001 (27). In addition, data from the Paffenbarger Physical Activity Index (24) collected in a subset of participants confirmed that ILI participants (n = 1112) had higher mean energy expenditure across the 8-year period than DSE (n = 1,086): 1041.14 versus 856.02 kcal/wk; p < .0001. Panel A of Figure 1 plots the mean (SE) % weight loss data at years 1, 4, and 8 for DSE versus ILI, whereas Panel B plots the mean (SE) energy expenditure in leisure-time physical activity for ILI versus DSE. Across all time points for both % weight loss and physical activity, ILI outperforms DSE (all p values < .0001) although there is evidence of weight regain and a reduction in physical activity in ILI from year 1 to 8.

Physical Function Across 8 Years of the Look AHEAD Interventions

Figure 2 plots PF by intervention group over time and illustrates that the ILI treatment led to a significant improvement in self-reported PF in the first year compared with DSE (p < .0001). Together with the overall intervention effect (p < .0001; see Table 2), the magnitude of improvement was approximately 2.64 units on the scale of self-reported PF. This effect of ILI was maintained across the 8 years of the study despite a significant decline in function with time among both ILI and DSE (p < .0001); the effect of the intervention on the linear trend across time was not significant (p = .273). To place these results in a clinical context, at year 8 an ILI participant with a PF score of 48.5 (the mean baseline SF score) would have a level of PF that was approximately 2.67 years younger than a person in DSE. These data illustrate that the ILI buffered the rate of secular decline in PF due to aging that was observed in DSE. The marked separation of the confidence bands in Figure 2 suggests that the persistent effect of ILI is beyond random chance over the entire course of the study. It is interesting to note, however, that the rate of decline with aging is greater than the average effect induced by the ILI—an approximate 4-point secular decline in self-reported PF over the course of 8 years compared with the 2.6-point buffer created by ILI.

Figure 2.

Self-reported physical functioning by intervention status over 8 years of study period. The dotted lines represent 95% confidence bands.

Table 2.

Result of Generalized Estimating Equation Analysis for the Intervention Effect

Variable	Estimate (95% CI)	p Value
Intervention	1.64 (1.14, 2.14)	<.0001
First year assessment	1.00 (0.68, 1.32)	<.0001
Time	−0.58 (−0.64, −0.55)	<.0001
Baseline physical functioning	0.69 (0.64, 0.72)	<.0001
Intervention × Time	−0.05 (−0.14, 0.04)	.273

Based on a division of the cohort at the median age at baseline (58.8 years), Figure 3 shows that, although both younger and older participants benefited from the ILI treatment compared with DSE, there was a greater benefit for older than younger participants (p = 0.02). Also, there was strong evidence that the rate of decline was steeper for older than younger participants with aging (p < .0001). To further elucidate the effect that ILI had on PF as participants aged, we created a plot of mean PF scores across the 8 years of follow-up as a function of age. This descriptive illustration, shown in Figure 4, reveals that the differential rates of decline between ILI and DSE are clearly evident at age 65 and progressively increase until age 75. There are limited data beyond age 75; however, the trend appears to hold until age 80.

Figure 3.

Plot of the SF-36 physical functioning subscale by treatment group for younger and older participants across 8 years of the Look AHEAD study.

Figure 4.

Plot of differential decline in the average SF-36 physical functioning subscale scores across 8 years by treatment group and age as a continuous variable.

Figure 5 shows that, although participants with and without CVD at baseline experienced a benefit from the ILI treatment at year 1, those with CVD had lower PF at baseline and had lost all benefit accrued from the ILI treatment by the 2nd year of follow-up assessment. Across the 8 years, there was no significant benefit of ILI treatment for participants with CVD (p = .43). Furthermore, although it appears that the rate of decline for participants with CVD across the 8 years was greater than for those without CVD, this difference was not statistically significant (p = .81), regardless of treatment (p = .12).

Figure 5.

Plot of the SF-36 physical functioning subscale by treatment group for persons with and without cardiovascular disease across 8 years of the Look AHEAD study.

Clinical Relevance of the SF-36 Measure of PF

Using baseline data, we examined the clinical relevance of the SF-36 PF scale. Scores on this measure had significant correlations with peak METs (r = .40, p < .001), kcal expended in physical activity (r = .19, p < .0001), and body mass index (r = −.31, p < .0001). When we partitioned the PF data by quartiles (Q1 = low functioning and Q4 = high functioning), the mean T scores for Q1–Q4 were 37.17, 47.95, 52.58, and 56.13, respectively. There was a higher percentage of women than men in Q1 (30.03% vs. 17.54%; p < .0001) and a lower percentage of women than men in Q4 (21.05% vs. 31.03%; p < .0001). Age was negatively correlated with PF scores (r = −.12, p < .0001), and there were more Blacks than Whites in Q1 (29.58% vs. 23.71%; p < .0001). More participants in Q1 self-reported that they had a history of hypertension, CVD, chronic obstructive pulmonary disease, or arthritis than those without these comorbidities, and there was a lower percentage of participants with these comorbidities in Q4 (see Table 3). Finally, we created three categories of diabetes severity (12): no diabetes-related drugs, oral diabetes drugs only, and any use of insulin. There was an increased percentage of participants in Q1 with increasing severity of diabetes—21.82%, 23.26%, and 32.87%; p < .0001—and a decreasing percentage in Q4—27.27%, 26.15%, and 19.44%—with those using insulin being at highest risk for having poor PF.

Table 3.

Percentages in Lowest and Highest Quartile on SF-36 Physical Functioning Subscale With Different Comorbidities

Comorbidity	Lowest Quartile—Q1		Highest Quartile—Q4
	Yes	No	Yes	No
Hypertension	26.83%	15.72%	23.16%	34.34%
Cardiovascular disease	35.06%	23.37%	16.59%	26.36%
Chronic obstructive pulmonary disease	40.89%	24.25%	11.11%	25.65%
Arthritis	36.97%	16.59%	13.83%	32.85%

Adverse Symptoms Related to Treatment

Throughout the study period, there was little evidence that ILI and DSE differed in physical symptoms related to the increased exercise behavior that occurred in ILI. There was a slightly higher incidence of pulled or strained muscles (p = .005) reported by participants in the ILI (18.7%) compared with DSE (15.6%) at year 1, which remained significant through year 3 (p = 0.018 at year 2 and p = 0.014 at year 3; see Supplementary Appendix, Table 1).

Discussion

This study provides the longest follow-up data to date on changes in self-reported PF among individuals with T2DM following an intensive weight loss intervention. Our analyses also provide strong support for the clinical utility of the SF-36 PF scale. As evident from the patterns of weight loss and physical activity across the 8 years, ILI achieved the desired intervention effect compared with DSE, albeit differences in weight loss and physical activity between ILI and DSE were substantially lower at year 8 than year 4. The challenges of long-term maintenance of weight loss and increased physical activity behavior have been well-documented (28–30). In light of the reduction in the ILI treatment effect over time for both weight loss and physical activity behavior, it is interesting to note that differences between ILI and DSE across the 8 years of the study for PF were relatively constant. This pattern in the data suggests that a legacy effect may be operative and warrants attention as this cohort is followed into the future.

Our results build upon existing publications of 4-year data from Look AHEAD which attest to the long-term efficacy of the ILI on weight loss, increased fitness, an improved CVD risk profile (27), and buffering the loss of mobility as people age (12). It also reinforces and extends the observation across 8 years that ILI improves general perceptions of physical health status compared with DSE (13); however, as reported in the primary outcome paper, there was no difference between ILI and DSE in hard CVD outcomes (16).

The current findings also reinforce and extend favorable results from other randomized clinical trials involving the study of weight loss and exercise on physical disability (30–32). Particularly promising was the moderating role that age had on the difference in PF between groups. That is, older participants experienced a larger benefit from the ILI than younger participants. Although this age difference may be due to the lower PF of the older participants at the time of baseline testing, both the Diabetes Prevention Program (14) and Look AHEAD (15) have found that older adults respond favorably to lifestyle interventions and lose more weight than younger participants.

Conversely, the presence of CVD at the time of the baseline assessment negatively influenced sustaining the benefit observed with ILI. That is, by the time participants with a history of CVD reached the second year follow-up assessment, their PF score was not statistically different from DSE. Although previous studies of lifestyle interventions in persons with CVD have observed positive effects on various measures of PF (33), these studies have been short in duration lasting less than 2 years. It is unclear why participants with CVD lost the initial, positive effect that ILI had on PF; however, this is a result that deserves further study and may be unique to persons with the combined disease burden of CVD and T2DM. Readers may also wonder whether this negative influence on the treatment effect was due to CVD per se or, differently, to the burden of an extra clinical condition. Although an extensive examination of moderator variables was beyond the scope and intent of this study, we did test a model in which OA was included as a comorbidity because it is well known that OA has a major effect on physical disability. When OA was entered into the model, it was significant, but did not alter the results of our analyses. Thus, we feel confident that, while CVD may not be the only moderator of the treatment effect, it is important one and particularly interesting given the results of the main outcomes paper (16).

The results of this investigation do have limitations. First, the outcome involved a single subjective measure, the PF subscale of the SF-36. Although this measure was found to have clinical utility, the ideal study would have included at least one objective measure of PF such as the 400-m walk test (32). Furthermore, it would be interesting to know whether the intervention improved limitations in any valued activities that participants were experiencing at the time of baseline testing. Second, the study sample was age restricted; it is unclear how adults 80 years or older and those with compromised function would have responded to treatment. In this regard, a short-term study by Villareal and colleagues (31) has shown that older, obese adults who have compromised physical health status do benefit from lifestyle interventions that include caloric restriction and exercise training.

In summary, an ILI was effective in improving self-reported PF compared with DSE in a group of overweight adults with T2DM for up to 8 years. However, this effect is not sustained in persons with a history of CVD. An ILI that targets weight loss and improved fitness might be a viable and clinically significant treatment for addressing the increasing Medicare costs associated with obesity and diabetes, particularly the adverse effect that these conditions have on falls, physical disability, and rates of conversion to major mobility disability as people age (34).

Supplementary Material

Supplementary material can be found at: http://biomedgerontology.oxfordjournals.org/.

Funding

This project was funded by the National Institutes of Health through cooperative agreements with the National Institute of Diabetes and Digestive and Kidney Diseases: DK57136, DK57149, DK56990, DK57177, DK57171, DK57151, DK57182, DK57131, DK57002, DK57078, DK57154, DK57178, DK57219, DK57008, DK57135, and DK56992. Additional funding was provided by the National Heart, Lung, and Blood Institute; National Institute of Nursing Research; National Center on Minority Health and Health Disparities; NIH Office of Research on Women’s Health; and the Centers for Disease Control and Prevention. This research was supported in part by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases. The Indian Health Service (I.H.S.) provided personnel, medical oversight, and use of facilities. The opinions expressed in this paper are those of the authors and do not necessarily reflect the views of the I.H.S. or other funding sources. Additional support was received from The Johns Hopkins Medical Institutions Bayview General Clinical Research Center (M01RR02719); the Massachusetts General Hospital Mallinckrodt General Clinical Research Center and the Massachusetts Institute of Technology General Clinical Research Center (M01RR01066); the University of Colorado Health Sciences Center General Clinical Research Center (M01RR00051) and Clinical Nutrition Research Unit (P30 DK48520); the University of Tennessee at Memphis General Clinical Research Center (M01RR0021140); the University of Pittsburgh General Clinical Research Center (GCRC) (M01RR000056), the Clinical Translational Research Center (CTRC) funded by the Clinical & Translational Science Award (UL1 RR 024153) and NIH grant (DK 046204); the VA Puget Sound Health Care System Medical Research Service, Department of Veterans Affairs; the Frederic C. Bartter General Clinical Research Center (M01RR01346); and a methodological grant awarded to E.H.I., R21AG042761-01. The following organizations have committed to make major contributions to Look AHEAD: FedEx Corporation; Health Management Resources; LifeScan, Inc., a Johnson & Johnson Company; OPTIFAST® of Nestle HealthCare Nutrition, Inc.; Hoffmann-La Roche Inc.; Abbott Nutrition; and Slim-Fast Brand of Unilever North America. Some of the information contained herein was derived from data provided by the Bureau of Vital Statistics, New York City Department of Health and Mental Hygiene.