- Click here for explanation of classification scheme.

Assess the effectiveness over two years of lifestyle intervention (i.e., diet, exercise or combination of diet and exercise) on changes in eating and exercise behavior, body weight, cardiovascular risk factors and incidence of diabetes in overweight individuals with parental history of diabetes.

• Overweight: 30 to 100% of ideal body weight (IBW)
• 40 to 55 years of age
• Non-diabetic (given OGTT to confirm)
• One or two biological parents with Type 2 diabetes
• Willing to sign consent approved by University of Pittsburgh IRB.

Diabetes.

Recruitment

Newspaper advertisements were used to recruit subjects. 

Design

Subjects were randomly assigned to one of four treatment conditions. 

Blinding Used

All scoring of three-day records was blinded to treatment condition and phase (baseline, six months or two years). 

Intervention

Treatment conditions

• Control
  • Subjects (N=40) were given copy of the LEARN Manual (self-help behavioral manual providing information on healthy eating, exercise and behavioral strategies for weight control) and encouraged to lose weight and exercise on their own
  • Subjects were not invited to treatment meetings but participated in all assessments 
• Diet
  • Subjects (N=37) attended weekly group meetings for first six months, then bi-weekly meetings for six months
  • Two six-week refresher courses were given during year two
  • Group meetings led by a behavior therapist and registered dietitian (RD) focused on behavioral strategies to help modify intake and nutrition information related to a low-calorie, low-fat regimen
  • Subjects follow 800 to 1,000kcal per day diet (20% kcal from fat) weeks one to eight, then diet was gradually made more flexible over the course of the program, with kcal goals adjusted to 1,200 to 1,500kcal per day at week 16
  • Meal plans and shopping lists were distributed weekly
  • Subjects self-monitored kcal and fat (g) intake every day for six months and periodically thereafter. Food records were reviewed weekly by the nutritionist who provided individual feedback.
  • At weekly sessions, the therapist presented information on a behavioral or nutritional topic.
     
• Exercise
  • Subjects (N=37) attended meetings on same contact schedule as diet condition. Meetings were presented by the behavior therapist and exercise physiologist. Each meeting included a lecture on a topic related to changing exercise behavior.
  • To facilitate adoption of an exercise routine, subjects took 50 to 60 minute walk with the therapist at each weekly meeting. During weeks one to 10, a second supervised walk session was available each week.
  • Subjects gradually increased physical activity to 1,500kcal per week through activities (i.e., brisk walking) completed five days per week and monitored exercise daily throughout program
  • Activity level increased gradually in bi-weekly increments of 250kcal per week, based on subject's self-reported activity level
  • Periodically, other activities (e.g., step aerobics and line dancing) were introduced.
     
• Diet and exercise
  • Subjects (N=40) received both diet and exercise interventions described above and encouraged to participate in supervised walks
  • Meetings were equal in frequency and length to diet or exercise condition. Thus, neither diet nor exercise received as much attention in this group as in diet-only or exercise-only groups. 

Statistical Analysis

• Statistical analysis done using  SAS and S-plus.
• One-way analysis of variance used to compare four treatment conditions on physiological measures at baseline and changes over time (six, 12 or 24 months minus baseline) 
• Behavioral variables were not normally distributed, thus non-parametric methods (Kruskal-Wallis) used to analyze these variables
• Three time points (six, 12 and 24 months) evaluated separately to allow for analysis of shorter- vs. longer-term changes
• Analysis completed only on subjects who attended the specific assessment; analyzed each time point separately to allow maximum sample to be used
• Analysis done for each treatment group to test differences across time. A four-group comparison was used in preference to 2X2 analysis because interventions were implemented differently depending on whether they were used alone or in combination.
• Partial proportional odds model used to predict disease state (normal, IGT or Type 2 diabetes) at year two
• Significance set at P<0.05. 

Timing of Measurements

• Baseline, six months, one year and two years
• All variables collected at each time point unless exception is noted below.

Dependent Variables

• Body weight: Used balance-beam scale with subjects weighed in hospital gowns without shoes
• BMI: Used stadiometer to measure height which was used with weight to compute BMI (kilograms per square meter)
• Waist and hip circumference: Measured at umbilicus and at level of maximal protrusion of buttocks, respectively, using Gulick tape measure
• OGTT: Administered at baseline and two years using 75g glucose load. Samples for glucose and insulin collected at zero, 30, 60, 90, 120 minutes.
• Fasting glucose: Assessed at year one; glucose levels analyzed using Beckman Glucose Analyzer and insulin analyzed by radioimmunoassay
• Glycosylated hemoglobin (HbA₁) measured on saline-incubated samples by column chromatography measuring the stable fraction
• Lipid measures analyzed in CDC approved lipid laboratory
  • Fasting cholesterol
  • Triglycerides
  • HDL cholesterol
• Blood pressure: Measured with random zero phygmomanometer with subjects seated by a five-minute rest period and duplicate measures obtained
• Exercise: Assessed via self-report using Paffenbarger Physical Activity Questionnaire via interview. Fitness was measured by subjects completing a half-mile walk test on an indoor measured track. Time to completion determined and regression formulas used to predict mzx oxygen uptake (VO₂ max).
• Dietary intake: Assessed via Block Food Frequency, modified to reflect a six-month interval. Questionnaire was self-administered and analyzed using National Cancer Institute Dietary Analysis software (4.01) and Nutrient and Portion software (versions 3.4). Three-day food diaries collected and documented by trained personnel and analyzed using Nutrition-Data-System software (version 2.5; Minneapolis, MN) released in 1993 (food database 7A; nutrient database S22). All scoring of three-day records done blinded to treatment condition and phase (baseline, six months or two years).

Independent Variables

Diet, exercise and combination of diet and exercise.

Control Variables

No-treatment control group.

 

• Initial N: 154 (79% female)
• Attrition:
  • 131 completed six-month assessment (85% of initial cohort)
  • 120 completed one-year assessment (78% of initial cohort)
  • 129 completed two-year assessment (84% of initial cohort)
• Age: 45.7±4.4 years
• Ethnicity: Not stated 
• Other relevant demographics: No significant differences between conditions seen for any baseline characteristics (all P-values>0.10)
• Anthropometrics: BMI of 35.9±4.3kg per m² at baseline 
• Location: Pittsburgh, Pennsylvania. 

 Response After Six Months of Intervention by Intervention Group

Measure	Control	Diet	Exercise	Diet and Exercise	P-value
N (percentage)	32 (80)	35 (95)	33 (89)	31 (78)	N/A
Diet (kcal/day)
Three-day Record	-332±690*	-692±525**	-92±724	-640±541**	<0.001
Block FFQ	-164±628	-707±777**	-275±836	-595±766**	<0.005
Percentage of Fat
Three-day Record	-2.1±7.4	-8.6±9.8**	-0.2±8.5	-11.3±9.5**	<0.001
Exercise
Paff (kcal per week)	89±948	178±825	657±897**	1,028±1,380**	<0.001
Walk Test (minutes)	0.1±0.6	-0.5±0.5**	-0.4±0.4**	-0.9±0.8**	<0.001
VO2 max (ml/kg per minute)	0.18±1.74	2.35±1.45**	1.09±1.16**	3.06±2.19**	<0.001
Weight/Fat Distribution
Weight (kg)	-1.5±2.7**	-9.1±6.4**	-2.1±4.2**	-10.3±7.7**	<0.001
BMI (kg/m2)	-0.6±1.0**	-3.3±2.2**	-0.8±1.5**	-3.7±2.6**	<0.001
Glucose/Insulin
FPG (mmol/l)	0.1±0.5	-0.2±0.4**	0.0±0.7	-0.2±0.4**	<0.05
Insulin*** (pmol/l)	2±45	-24±52**	-4±52	-37±112*	<0.05
Lipids/Blood Pressure
Chol (mmol/l)	0.12±0.50	-0.49±0.71**	0.12±0.72	-0.33±0.61**	<0.001
HDL (mmol/l)	-0.2±0.11	-0.10±0.17**	0.02±0.16	-0.06±0.16	<0.05
LDL (mmol/l)	0.08±0.46	-0.32±0.60**	0.03±0.52	-0.13±0.58	<0.05
TG ***(mmol/l)	0.2±1.32	-0.30±1.45	0.12±0.64	-0.69±1.45**	<0.05
Systolic BP (mmHg)	-2.0±10.5	-10.2±9.2**	-2.4±18.9	-12.3±9.5**	<0.005

Data are means±SD. *P<0.05; **P<0.01 for changes from baseline; ***Log-transformed for analysis. P-values for four-group comparisons from Kruskal-Wallis tests for behavioral measures and from analysis of variance for physiological measures.

Response After One Year of Intervention by Intervention Group

Data are means ± SD. *P<0.05; **P<0.01 for changes from baseline. P-values for four-group comparisons from Kruskal-Wallis tests for behavioral measures and from analysis of variance for physiological measures.  

Response After Two Years of Intervention by Intervention Group 

Measure	Control	Diet	Exercise	Diet and Exercise	P-value
N (percentage)	31 (78)	35 (95)	31 (84)	32 (80)	N/A
Diet (kcal/day)
Block FFQ	-49±500	-521±597**	-206±771	-325±800	<0.05
Glucose/Insulin
Insulin *** (pmol/l)	1±53	20±50	43	-12±89	<0.05
Lipids
Cholesterol (mmol/l)	0.18±0.53	-0.12±0.61	0.33±0.64**	0.09±0.67	<0.05

Data are means ±SD. *P<0.05; **P<0.01 for changes from baseline; ***Log-transformed for analysis. P-values for four-group comparisons from Kruskal-Wallis tests for behavioral measures and from analysis of variance for physiological measures.

Other Findings

• Baseline: No significant differences between treatment conditions seen for any variable (all P-values>0.10)
• Six months: 
  • Groups differed significantly on measures of eating, exercise and fitness
  • Weight losses in diet and diet plus exercise groups were significantly greater than exercise and control conditions
  • Physiological changes were also significantly different by treatment condition, with diet and diet and exercise condition losing the most weight (9.1kg and 10.3kg, respectively)
  • Weight changes in exercise (-2.1kg) and control (-1.5kg) conditions were also significantly different from zero, but far more modest
  • Significant differences between conditions also seen for FPG and insulin, total cholesterol, HDL, LDL, TG and systolic BP
  • Changes in these variables were greatest in diet and diet and exercise conditions.
• One year:
  • Attendance at sessions from six to 12 months dramatically reduced compared with initial six months, and differences between conditions were significant (P<0.02)
  • Diet group attended 37%, exercise group attended 16% and diet and exercise attended 29% of their sessions
  • Behavioral measures collected at month 12 suggest that changes in diet and exercise were being maintained in the relevant treatment groups. However, weight loss at one year decreased in all groups. 
  • The diet and diet and exercise conditions continued to have significantly better weight loss than the exercise or control groups (P<0.0001)
  • There were no significant differences between conditions on any of the other physiological parameters at 12 months.
• Two years:
  • Two refresher courses were offered during the second year. Attendance at these courses differed by condition (P<0.005), with subjects in diet condition attending 36%, exercise condition attending 15% and diet and exercise condition attending 29%. Attendance at the two-year assessment did not differ significantly by treatment group.
  • Neither behavior changes, nor physiological changes were well-maintained at two years
  • The four conditions no longer differed in Paffenbarger activity scores or either of the fitness measures
  • Groups did not differ significantly in dietary fat intake, despite continued lower overall energy intake in diet and diet and exercise conditions
  • Weight loss did not differ by treatment condition
  • Only the diet and exercise group maintained a significant decrease in body weight (-2.5kg; P<0.01) from baseline to two years
  • Likewise, percentage of subjects maintaining clinically significant weight loss at two years did not differ by treatment group (X²=6.8; P=0.08)
  • Physiological variables showed no sustained improvements
  • FPG was significantly above baseline in all four conditions, indicating a worsening in glycemic control with time
  • The only significant or borderline significant differences between groups were LDL, total cholesterol and insulin, where increases were observed in exercise condition and slight improvements (in lipid levels) in diet conditions
  • Weight loss at two years was related to estimated VO₂ max (r=-0.44; P<0.0001), walk time (r=0.29; P<0.002), calorie intake (r=0.27; P<0.005) and percentage of kcal as fat (r=0.27; P<0.005) at two years
  • Differences between groups in risk of developing diabetes were borderline significant (P=0.08)
  • Strongest predictors related to risk of developing diabetes:
    • Impaired glucose tolerance at baseline was positively related to risk (P<0.0001)
    • Weight loss from baseline to two years was negatively related to risk (P<0.003)
  • In all treatment groups, a modest weight loss of 4.5kg reduced risk of Type 2 diabetes by approximately 30% compared with no weight loss.

 

Although initially successful, the interventions studied were not effective in producing long-term changes in behavior, weight or physiological parameters. However, weight loss from zero to two years reduced risk of developing Type 2 diabetes. 

Government:

NIH, GCRC, NIDDK

Exclusion criteria not indicated.