NC: Weight Management (2007-2008)


Dornelas EA, Wylie-Rosett J, Swencionis C. The DIET study: long-term outcomes of a cognitive-behavioral weight-control intervention in independent-living elders. Dietary Intervention: Evaluation of Technology. J Am Diet Assoc. 1998 Nov; 98 (11): 1,276-1,281.

Study Design:
Before-After Study
D - Click here for explanation of classification scheme.
Quality Rating:
Neutral NEUTRAL: See Quality Criteria Checklist below.
Research Purpose:

To determine if weight loss and improvement in cardiovascular risk status, derived from a cognitive-behavioral weight-control intervention at 40 weeks, was maintained long-term (two to three years).

Inclusion Criteria:
  • Residents of two different independent living retirement communities
  • Overweight (subjects were at least 4.5kg overweight, using Age-Specific Gerontology Research Center Height-Weight Tables); specific weight criteria for inclusion was not given.
Exclusion Criteria:

None stated.

Description of Study Protocol:


  • Subjects were recruited from two different independent living retirement communitites
  • All subjects were overweight.


  • Quasi-experimental
  • This study evaluates subjects from a previous study to see if weight-loss and decrease in cardiovascular risk factors were maintained for two to three years
  • The original study compared two groups; the intervention group was compared with a wait-listed control group at 40 weeks after baseline
  • The controlled trial ended at 40 weeks, then both groups received two years of intervention
  • This study evaluates all of the subjects after two years
  • The original intervention group was evaluated at three years, one year after all intervention ended.


"DIET" (Dietary Intervention: Evaluation of Technology)

  • 10 weeks of an intensive psycho-educational approach focused on lifestyle change
  • Followed by a less-intensive phase focusing on relapse prevention and maintenance of lifestyle changes
  • Program designed to optimize use of staff time (registered dietitians and health psychologists)
  • During the first 10 weeks, participants kept food and exercise diaries, completed workbook assignments, received weekly videotapes and attended weekly weigh-ins
    • At baseline, a 15-minute individual counseling session with a health psychologist or a registered dietitian was conducted to set goals and review cardiovascular risk factors
    • Clerical staff weighed participants and recorded whether diary and workbook chapters were complete
    • Clerical staff was available two to three mornings per week and professional staff were available once per week
    • Additional counseling sessions were shceduled with professional staff if participants had problems with the program or with achieving goals
  • During the following 22 months, participants attended monthly weigh-ins, completed food and exercise diaries and were provided professional counseling as needed
  • Four workshops were held, six weeks apart and focused on relapse prevention, physical activity, dietary fat reduction and eating behavior. 

Statistical Analysis

  • Statistical Package for Social Sciences (6.1.3, 1995 Chicago, IL)
  • Comparison of categoric and continuous baseline variables for all groups were performed using chi squared and one-way analysis of variance, respectively
  • Bonferroni test was used to adjust for inflation, due to multiple comparisons of groups on baseline data
  • Trend analysis, using a within-subjects repeated-measures analysis of variance design that tested for significant changes in weight and blood values over time
  • Orthogonal polynomial coefficients for each time-point were contrasted to detect linear trends in weight and blood values over time.
Data Collection Summary:

Timing of Measurements

  • Baseline
  • End of 40-week intervention
  • Two years after baseline for all subjects
  • One year after all intervention ended for the initial intervention group.

Dependent Variables

  • Variable One: Lipid levels (HDL-C reported), measured in a lab that is certified by the state of New York Dept. of Health, determined in a m anner consistent with well-established criteria
  • Variable Two: Plasma insulin level, measured in duplicate by radioimmunoassay
  • Variable Three: Serum glucose level, determined byhexokinase method, using coupled enzyme reactions catalyzed by heokinase and glucose-6 phosphate dehydrogenase
  • Variable Four
    • Weight was measured in the morning with participants in light indoor clothing and no shoes
    • Standard physician beam-balance scales were calibrated daily using an 11.5-kg weight
    • Weight was recorded to the nearst 0.11kg. 
  • Variable Five: BMI was calculated using weight (kg) divided by height (m) squared
    • Height was measured in the same clothes, no shoes as weight
    • Equipment for measuring height was not given.
  • Variable Six: Psychological well-being was assessed using the psychological General Well-Being Schedule
  • Variable Seven: Walking (blocks or time per day).

Independent Variables

  • Initial intervention group or control community
  • Each participant baseline to end of study.

Control Variables

  • Variable One: Sex; men (N=32) women (N=141)
  • Variable Two: Age
  • Variable Three: Education level
  • Variable Four: Smoker or non-smoker
  • Variable Five: Chest pain on exertion
  • Variable Six: Hypertension (self-report)
  • Variable Seven: Diabetes (self-report).
Description of Actual Data Sample:

Initial N

  • 325 residents signed informed consent and agreed to participate in study (at baseline)
  • 247 participants started the program at baseline
  • Number of subjects still in the program at 40 weeks (when this phase of the study started) is not given.

Attrition (Final N)

  • 173 participants completed the two years (84 from the original intervention group; 89 out of the original control group)
  • 63 of the original intervention community were available for measurements at three years post-baseline
  • Reasons for attrition not given, except that four people who started the program died before the study was complete.


71.2 years (SD=6.6).


100% Caucasian.

Other Relevant Demographics

  • All participants were recruited from two independent living retirement communities
  • 66% were currently married.


Paricipants were at least 4.5kg overweight.


State of New York.

Summary of Results:



Measures and Confidence Intervals)

2 Years
(Measures and Confidence Intervals)

Statistical Significance of Group Difference

3 Years; Original Intervention Group, 1 Year After All Intervention Stopped
(Measures and Confidence Intervals)

Statistical Significance of Difference
Dependent Variable 1 Mean, CI Mean, CI      
  Lipid levels (HDL-C reported) 1.2±0.35mmol/L= 46.4±13.5mg/dL 1.2±0.32mmol/L= 46.4±12.4mg/dL Stat. signif. diff. between groups    

Percentage under 1.04 mmol/L = 40.2mg/dL



Not significant

1.2±0.04mmol/L= 46.4±1.5mg/dL

Not significant

Dependent Variable 2












Serum glucose level


5.8±1.3mmol/L =104.4±23.4mg/dL


P<0.5 5.2±0.26mmol/L= 93.6 ±4.68mg/dL  
  Glucose over 5.5mmol/L (99mg/dL). Participants with self-reported diabetes (N=13) excluded 45% 22%     Significance not reported.  "Decreases from baseline were maintained."
Dependent Variable 3          





Not given

Dependent Variable 4          





Baseline for just the 63 participants measured at 3 years =30.9±4.5kg/m2

3 yr f/u:  29.5kg/m2±4.7


Not significant between 2 and 3 years (P+0.09)

Dependent Variable 5          

Psychological well-being (higher scores reflect lack of psychological distress and presence of subjective well-being) 



Not significant



Dependent Variable 6






Walking (blocks or time per day)

9.3±10.8 blocks 

13.1± 12.7 blocks


P<0.05 12.1±11.9 blocks  Difference from 40 weeks  and 2 years not significant
  Walking over 20 minutes per day 45% 55% Difference between 40 weeks and 2 years not significant    

Other Findings

  • As a group, 70% of the completers lost some weight by two years (0.68kg to 23.3kg; mean+SD=2.7±4.6kg). Men lost more weight than women. There was a significant within-subject effect on decrease in BMI (P=0.001).
  • Baseline insulin levels of fasting subjects were modestly correlated with baseline BMI (R=0.26, P=0.05), moderately correlated with BMI at 40 weeks (R=0.47, P=0.000) and modestly correlated with BMI at two years (R=0.27, P=0.06)
  • Time spent walking, rather than number of blocks walked daily, was predictive of HDL-cholesterol level at the three-year follow-up (P=0.004): Participants who walked 20 to 30 minutes and more than 30 minutes had higher HDL-cholesterol levels  (1.3±0.36mmol per Land 1.5±0.20mmol per L, respectively) compared with those who walked shorter times
  • 54 out of 74 people who signed consent forms but did not start the program completed baseline questionaires, but did not have any blood work done. These "non-starters" weighed less than both those who completed the program and those who started the program and later dropped out. People who started the program but did not complete it were heavier and had a consistent trend toward poorer perceived health and less physical activity.


Author Conclusion:
  • This community-based program achieved long-term weight loss, lowered fasting blood glucose level and achieved lifestyle change by increasing physical activity in a community of older adults
  • Community-based intervention programs that optimize use of professional staff time can be used to lower body weight and improve blood glucose levels in older adults and these improvements to the cardiovascular risk profile appear to endure for the long-term
  • Programs of this type could be offered in retirement communitites or in combination with a worksite health education program
  • Suggestions for modifying this intervention include assessing readiness to change and targeting the intervention using a stage-based approach.

The author reports the study had two major limitations

  1. The "quasi-experimental model," with use of a wait-listed control group: This precluded the collection of control data at the two- and three-year follow-up points. Due to the community based nature of this program, keeping people on a waiting list for a longer time was not feasible.
  2. The three-year assessment of the 63-person intervention community follow-up group indicated that participants were significantly heavier and more likely to have hypertension at baseline than the community as a whole, who completed the two-year assessment. However, among the entire sample, non-completers (N=74) were significantly heavier than non-starters, but completers (N=173) did not differ from either group in regard to baseline weight. Given this finding, it seems probable that the higher baseline weight and the proportion of persons with hypertension were characteristic of the initial intervention community, rather than being a self-selection bias of those who completed the program.
Funding Source:
Government: NIH, NHLBI, NIDDK
Reviewer Comments:
  1. This study was intriguing in that it evaluated a long-term weight management and lifestyle program and showed that changes were maintained up to one year after intervention was stopped
    • The limitations in the reviewer's mind include the very large drop-out rate: 23% of people originally recruited did not start the program and 30% of those who started the program dropped out before it was completed. Reasons for attrition were not given.
    • Due to the nature of the program, it could not be randomized
    • There may have been a difference in the motivation level of the initial intervention group and the controls who were on a waiting list for the program
    • The fact that the study was not randomized and not blinded  in any way reduces its validity.
  2. Statistical analysis-SD very close to the mean, which indicates that the data are relatively widely dispersed and the mean is a poor representation of the full data set.