Effects of Walking Training on Weight Maintenance After a Very-Low-Energy Diet in Premenopausal Obese Women:  A Randomized Controlled Trial FREE

OBESITY HAS become a worldwide epidemic.¹ Several comorbidities are associated with obesity, such as type 2 diabetes mellitus, hypertension, coronary artery disease, osteoarthritis (knee and hip), and certain types of cancer. Weight reduction decreases the risk for insulin resistance and atherosclerosis through favorable changes in blood pressure, blood lipid levels, lipoprotein levels, and insulin levels.^1- 3

Although weight reduction is achievable by several methods, long-term maintenance of weight loss remains a challenge. A recent systematic review on treatment of obesity⁴ shows that most studies contained 1-year follow-up data after the intervention. Although the entire weight loss was usually not maintained, all but 1 study reported that the mean body weight was still at least 2 kg below pretreatment levels at follow-up. Results on long-term (at least 2 years) weight maintenance were more contradictory and less promising. In many studies, the subjects' mean body weight reached or went beyond pretreatment levels at 2 to 5 years' follow-up.^5- 9 In contrast, some studies^10- 12 reported an almost complete long-term maintenance of weight loss (judged from group mean values).

Factors associated with successful maintenance of weight loss have been retrospectively identified in many studies. Two factors most often cited are high levels of physical activity,^7,13- 16 and strong cognitive control over eating.^8,13,17 However, controlled trials with interventions on physical activity either during^14,18 or after^19- 21 weight reduction did not show any beneficial effects on long-term weight maintenance. The long-term effect of physical activity on the preservation of the fat-free body mass may be favorable.^14,22 Nevertheless, it is still unclear whether physical activity as a component of a weight reduction or a maintenance program has additional long-term effects on weight maintenance and obesity-associated disorders, such as blood lipid levels or insulin sensitivity.^{14,18,23- 24}

This study examined the role of physical activity in weight maintenance. The hypotheses were (1) a walking training program of moderate intensity, started after weight reduction by a very-low-energy diet, improves maintenance of weight loss and obesity-related metabolic disorders, and (2) the effect of the training program is related to the prescribed amount of physical activity, that is, a higher amount (energy expenditure) leads to more favorable results. The main outcome variables were body weight, fat mass, and waist circumference, assessed 2 years after the end of the maintenance program. Secondary outcome variables were blood pressure, and concentrations of total cholesterol, high-density lipoprotein cholesterol (HDL-C) level, triglyceride levels, glucose level, and insulin level in serum or plasma.

This 3-year study consisted of 3 phases. First, all subjects participated in weight reduction for 12 weeks. Subsequently, the subjects were randomized into the following 3 groups for 40 weeks: a control group with no increase in habitual exercise; a walk-1 group with a walking program targeted to expend 4.2 MJ/wk; and a walk-2 group, with a walking program of 8.4 MJ/wk. Random permuted blocks within strata were used,²⁵ with weight loss (in 3 classes) as the stratifying factor. The third phase was an unsupervised 2-year follow-up. If not otherwise specified, the subjects were assessed before and after the weight reduction, after the maintenance program, and at 1 and 2 years of follow-up.

A sample size of 25 participants was projected to detect a difference of 4.6 kg in weight change between the study groups, with the risk for type I error set at 0.05 and the risk for type II error set at 0.1.²⁶ The a priori estimated SD for weight change was 5 kg.

We received 207 letters from female volunteers responding to newspaper advertisements (Figure 1). The requirements in the newspaper advertisements were aged between 30 and 45 years, markedly overweight, and no illnesses or taking medication. Of the answers, 82 were rejected (mostly because of too low a weight, too old, or self-withdrawal), and 125 women were invited to screening assessments including medical history, physician examination, and submaximal exercise test.

The inclusion criteria to the study were being female, a body mass index ([BMI], weight, in kilograms, divided by height, in meters, squared) between 30 and 45, aged 30 to 45 years, premenopausal, clinically healthy, not regularly using medication (other than hormonal contraceptives). In addition, the weight had to be stable (±3 kg for at least the previous 3 months) and the volunteers ought not to be physically active (leisure-time physical exercise ≤2 times per week), pregnant, lactating, or smokers. Suspected binge eaters (symptom scores >20 on the Bulimic Investigatory Test of Edinburgh [BITE]²⁷) were excluded. In screening assessments, 40 women were rejected (mostly because of too high a BITE score, illnesses, or improper BMI). Thus, 85 women (mean BMI, 34.0; mean age, 40 years) were accepted. The study was approved by the Ethical Committee of the UKK Institute for Health Promotion Research. A written informed consent was obtained from the participants.

Eighty-two subjects completed the weight reduction phase; 80 subjects finished the maintenance program; 75 subjects were measured at the end of the 1-year follow-up; and 74 subjects were measured at the end of the 2-year follow-up (Figure 1).

The 12-week weight reduction consisted of 3 parts: week 1, low-energy diet based on a meal-exchange system; weeks 2 to 9, very-low-energy diet (Nutrilett; Nycomed-Pharma AS, Oslo, Norway); and weeks 10 to 12, low-energy diet.²⁸ The subjects met weekly in small groups (5-12 participants). All meetings were overseen by a nutritionist. The meeting topics included instructions on the diets and weight maintenance, and on relapse prevention.²⁹ All subjects were weighed before every meeting.

The intensity of walking exercise was set at 50% to 60% of individual heart rate reserve (maximal minus resting heart rate) added to resting heart rate. The calculation of the weekly walking time needed to cover the target energy expenditure during walking was based on a linear regression of heart rate vs oxygen consumption during a maximal exercise test (uphill treadmill walking). The time was calculated as the target energy expenditure divided by the energy expenditure (kilojoules per minute) during the exercise test corresponding to the target heart rate zone. On the average, the walk-1 subjects were prescribed to walk 2 to 3 hours weekly, and the walk-2 subjects, 4 to 6 hours weekly. The subjects used a heart rate monitor (Polar Edge; Polar Electro Oy, Kempele, Finland) during the walking sessions. One weekly walking session was supervised.

All subjects participated in weekly meetings in small groups throughout the maintenance program, conducted by an exercise instructor. All subjects were instructed to follow a low-fat diet, and they received educational material monthly. As a part of their homework, they were asked to monitor high-risk situations for overeating.²⁹ Problems in diet and prevention of relapses were discussed in the meetings.

Body weight was measured after an overnight fast, with a high-precision scale (F150S-D2; Sartorius, Goettingen, Germany), with the subjects wearing only their underwear. The subjects' body density was measured by underwater weighing, after full exhalation (presumably at residual lung volume), as described previously.²⁸ Body composition was calculated from the body density by a 2-component model.³⁰ Waist circumference was measured midway between the lowest rib and the iliac crest. The mean of 3 readings was used.

The metabolic (insulin-resistance) syndrome³¹ was diagnosed if at least 3 of the following characteristics were fulfilled (modified after Vanhala et al³²): BMI greater than 30, waist circumference greater than 90 cm, systolic blood pressure higher than 160 mm Hg and/or diastolic blood pressure higher than 95 mm Hg, HDL-C less than 1.1 mmol/L (42 mg/dL), triglyceride levels higher than 2.0 mmol/L (175 mg/dL), plasma glucose level of 7.0 mmol/L or more (≥126 mg/dL), and plasma insulin level greater than 104 pmol/L.

Blood pressure was measured in the sitting position using a random 0 sphygmomanometer. The mean of 2 measurements was used. Venous blood samples were obtained after a 15-minute rest while the patient was supine at 8 to 9 AM after a 12-hour fast. Serum was separated by centrifugation and stored at −70°C until analyzed.

Serum cholesterol and triglyceride concentrations were analyzed from frozen samples by enzymatic methods (CHOD-PAP for cholesterol and GPO-PAP for triglycerides; Boehringer Mannheim, Mannheim, Germany). High-density lipoprotein cholesterol level was determined by selective precipitation (dextran sulfate-magnesium chloride).³³ Plasma insulin determinations were done by radioimmunoassay (Phadeseph Insulin; Pharmacia, Uppsala, Sweden). Plasma glucose was assessed by the glucose dehydrogenase method (Merck Diagnostica, Darmstadt, Germany). An automatic analyser (Hitachi model 717; Hitachi Ltd, Tokyo, Japan) and analyser model (Epos 5060; Eppendorf, Darmstadt, Germany) were used for the analyses.

Each subject wore a pedometer (Fitty3; Kasper & Richter, Uttenreuth, Germany), which recorded the number of steps taken daily, for 7 days during week 7 of weight reduction, during the last week of the maintenance program, and at 1 and 2 years of follow-up.

Maximal oxygen consumption ([UNK]O₂max) was determined during an uphill walk on the treadmill until the symptom-free volitional maximum was reached.³⁴ Expiratory gases were analyzed using a sensor (Sensor Medics model 3000Z; Sensor Medics Corp, Anaheim, Calif) and integrated for 1-minute periods.

The Three-Factor Eating Questionnaire, which measures cognitive restrained eating, disinhibition (poor control over eating), and hunger,³⁵ was used before and after weight reduction, in the middle and at the end of the weight maintenance program (the mean of these 2 time points was used), and at 1 and 2 years of follow-up.

The group differences for the main and secondary outcomes were estimated by analysis of covariance, with the measurements after the weight maintenance program and at the 2-year follow-up separately as dependent variables. The respective measurement after weight reduction (before randomization) was used as a covariate. There were several reasons for using univariate analysis for each time point instead of analysis of covariance with repeated measurements. First, the last assessment point was regarded as more important than that after weight reduction. Second, the assessment points were preceded by different protocols (supervised vs unsupervised). Finally, the 2 assessment points were separated by considerably long periods. The differences of both exercise groups vs the control group were estimated using simple contrasts.

The effects of group, time, and group × time interaction on variables describing adherence to exercise and diet (the number of daily steps, [UNK]O₂max, indexes of eating control) were analyzed by analysis of variance, using data from before weight reduction, at the beginning and end of the maintenance program, and at the end of the follow-up. The group differences in proportions of subjects with the metabolic syndrome were analyzed by the method of generalized estimating equations.³⁶ The proportion was measured among those 65 subjects who had complete data for the components of the metabolic syndrome throughout the study.

All results are expressed as means (SDs) and 95% confidence intervals (CIs) for the difference (against the control group). P<.05 was used as the level of statistical significance. The statistical analyses were done using the SPSS statistical software package (version 8.0; SPSS Inc, Chicago, Ill). The generalized estimating equations models were done by S-Plus software package (version 4.0; MathSoft Inc, Seattle, Wash) with Oswald software library for analysis of longitudinal data (version 3.0; David M. Smith, Lancaster University, Lancaster, England).

All accepted subjects had a BMI in the range of 30 to 45 in the screening assessment. At the subsequent baseline assessment (about 2 weeks later, at the start of weight reduction), BMI of 4 subjects was between 29.0 and 29.9 and in 1 subject was above 45.0. Seventy-eight of the subjects had a BMI between 30.0 and 39.9. Waist circumference of all subjects was at least 87.0 cm (Table 1).

The mean weight loss after the 12-wk weight reduction phase was 13.1 kg (SD 3.5; weight reduction range, 4.5-20.8 kg). Also fat mass and waist circumference decreased. The means of the anthropometric variables were statistically significantly lower in the beginning of the maintenance program compared with those before weight reduction (Table 1 and Table 2).

During the maintenance program, the mean body weight increased by 2.0 kg in the control group, whereas the exercise groups' weight remained rather stable (the mean decrease by groups was 0.6-0.7 kg [4.3-4.9 kg] by groups) (Table 2). The weight change was 2.7 kg (95% CI, 0.2-5.2) less in the walk-1 group and 2.6 kg (95% CI, 0.0-5.1) less in the walk-2 group than in the control group (analysis of covariance [ANCOVA], P = .06). A similar pattern was observed for fat mass and waist circumference (ANCOVA, P = .18 and P = .08, respectively).

At the end of the follow-up, the mean weight increase by groups was 5.9 to 9.6 kg when compared with the start of the maintenance program. The SDs were 5.2 to 6.6 kg, which were somewhat larger than in the power calculations. The ANCOVAs showed marginal differences between the groups (P = .07-.11). The tests of contrasts showed that the walk-1 group had gained 3.5 kg (95% CI, 0.2-6.8) less weight and 3.8 cm (0.3-7.3) less waist circumference than the control group, whereas the changes in walk-2 were not statistically significantly different from those observed in the control group.

The total duration of the weight maintenance program and the follow-up was 33 months. During this time, 9 participants lost weight or did not gain more than 1 kg. Two participants regained more than 20 kg.

The indicators for the metabolic syndrome improved during the weight reduction, except for diastolic blood pressure which was unchanged (Table 1 and Table 3). During the maintenance program, despite a mean stable weight, the values of most indicators worsened, except for unchanged diastolic blood pressure and glucose level and increased HDL-C. At the end of the follow-up, the mean values of systolic blood pressure and glucose level were unfavorably higher than before weight reduction, while the observed increase in HDL-C was favorable for coronary risk. The group differences were not statistically significant at the end of weight maintenance or at follow-up (ANCOVA, P = .12-.94). The lowest P values were associated with a tendency of the walk-1 group to have a higher mean in HDL-C concentrations (0.10 mmol/L [0.002 mg/dL]; 95% CI, 0.00-0.20) and of the walk-2 group to have lower insulin concentrations (−8 pmol/L; 95% CI, −16 to 0) when compared with the controls after the maintenance program.

Before weight reduction, the number of subjects with at least 3 pathologic characteristics for the metabolic syndrome was 27 (of the 65 subjects having complete data at all assessment points). The number decreased to 11 after weight reduction (Table 4). During the weight maintenance program and the follow-up, the proportion started to increase, especially in the control group. The group difference in the proportion during the study was statistically significant (generalized estimating equations model, P<.05) between the walk-2 and control groups at the end of the follow-up.

When comparing the number of daily steps between different assessment points, the group × time interaction was statistically significant (P = .02). At the end of the maintenance program, the walk-2 group took 3040 (95% CI, 1110-4970) more steps than the control group (Figure 2). At the 1-year follow-up, the walk-1 group showed a higher (25,700 steps; 95% CI, 770-4370) step count than the controls. At the end of the follow-up, all 3 groups' number of steps was similar. Maximal oxygen consumption, expressed as liters per minute, or divided by the fat-free body weight, or changes in [UNK]O₂max, was not different between the groups during the weight maintenance or during follow-up (group × time interaction, P = .47; data not shown).

Dietary restraint and disinhibition scores showed a similar pattern in all groups (group × time interaction, P>.27). The mean score of dietary restraint was 15.5 (95% CI, 14.6-16.4) after weight reduction, remained at the same level (15.6; 95% CI, 14.8-16.3) during the maintenance program, but decreased to 12.1 (95% CI, 11.1-13.1) at the end of the follow-up. The mean final score was still higher than before weight reduction (mean, 8.8). The mean score of disinhibition was 6.6 (95% CI, 6.1-7.5) after weight reduction. It remained at the same level during the maintenance program (6.5; 95% CI, 5.8-7.2), but increased to 7.5 (95% CI, 6.8-8.2) at the end of the follow-up.

The results partially supported the hypothesis that a program of walking training prescribed after weight reduction improved maintenance of losses in weight, fat mass, and waist circumference. The unique finding was that the favorable effects were only observed in those women walking a moderate amount (2-3 h/wk). In contrast, subjects with a higher training target (4-6 h/wk) of the same moderate intensity did not maintain their weight, body fat, or waist circumference any better than the controls who received only dietary counseling.

Similarly to our design, Perri et al^19- 20 used a weight maintenance program after weight reduction. Their results did not indicate that aerobic exercise was better than other programs (extended therapist contact, social influence maintenance program, or a multifaceted program). However, these studies did not compare different exercise programs.

Some groups have studied the effects on weight maintenance of exercise programs prescribed at the beginning of weight reduction. Perri et al³⁷ showed that a home-based exercise program for 6 months led to better weight maintenance at 15 months, compared with subjects participating in supervised group sessions. Andersen et al¹⁶ found a tendency for better 12-month weight maintenance in women in a lifestyle activity group, compared with a group of structured exercise. In contrast, Weinstock et al¹⁸ did not find any differences in 22-month weight maintenance between women randomized into groups of diet alone, diet and aerobic training, or diet and strength training for 11 months. The above studies compared different modes of exercise, but none used different amounts of training with the same mode and intensity as in our study. We expected that the walk-2 group would have been more successful than the walk-1 group with only half the amount of prescribed energy expenditure.

A central issue in the effectiveness of a physical activity program is adherence, which is known to be a problem particularly in long-term interventions and among obese subjects.³⁸ We tried to increase the adherence by use of a very flexible program. The subjects agreed to a specific weekly duration of exercise at a given intensity level (measured by heart rate). It was possible to increase or decrease intensity by "trading" with decreased or increased duration; only the estimated weekly energy expenditure during exercise was fixed. More subjects requested a reduction, rather than an increase, of walking intensity. The subjects reported their weekly physical activity to the exercise instructor. Concurrently, specific barriers to exercise were addressed. Time constraints owing to work and illnesses were the most often reported reasons for adherence problems.

In our study, we found better weight maintenance in the walking group with a moderate amount of training (walk-1), who also showed the best exercise adherence (as reflected by the number of daily steps) 1 year after the maintenance program. The poor long-term adherence in the walk-2 group may be related to a too time-consuming program. However, the higher number of steps in the walk-1 group was only seen at the 1-year follow-up.

Perri et al³⁷ showed better long-term (15 months) adherence in subjects who exercised according to a home-based program, compared with subjects in structured group-based exercise. Moreover, Weyer et al³⁸ found that obese subjects preferred unstructured lifestyle activity, rather than structured exercise. Those and the present results suggest that the key issues in exercise prescription for obese subjects are moderate intensity, moderate volume, and individuality.

In accord with earlier studies,^18- 20 weight was not totally maintained in any of our study groups during the unsupervised follow-up. In fact, Perri et al²⁰ have repeatedly shown that a maintenance program is effective only as long as the active intervention is ongoing. Unfortunately, when the external influence is removed, weight regain is usually unavoidable,^19- 20 a phenomenon also seen in our study. It is likely that extended therapist contact and support are the most important factors behind good weight maintenance.

Our subjects showed desirable cognitive (high dietary restraint, low disinhibition) and behavioral (higher physical activity) changes during the maintenance program. However, such changes needed to maintain energy balance were gradually lost during the follow-up. Therefore, weight regain was expected.

It has been suggested that physical activity has an additional decreasing effect on fat mass^14,22 and abdominal fat,³⁹ independent of the effects on weight loss. This was not found in our study, since fat mass and waist circumference followed the same pattern of change as total body weight. It is possible that the training intensity needs to be higher, before effects on body composition are seen.¹⁶ Males may also be more responsive to exercise than females.⁴⁰

Previous studies have not shown that physical activity, without a substantial weight loss, improved metabolic disorders associated with obesity. The favorable effects of weight reduction on hypertension and lipid and carbohydrate disorders are clear, but the additional effects of physical activity seem modest^23,41- 42 or not observable at all.^18,43 Despite stable mean weight, the indicators for metabolic disorders showed undesirable changes during the weight maintenance program in our study. This was in agreement with other studies on premenopausal women.^16,18 Some of the effects seen during weight loss were apparently acute responses to greatly reduced dietary (especially fat) intake. The increase in HDL-C concentration may be attributable to an increase in the activity of adipose tissue lipoprotein lipase (which transfers lipids to HDL-C) and a decrease in the activity of hepatic lipase (which removes HDL-C the blood).¹⁴ The increase in HDL-C is enhanced in the weight-stable phase after weight reduction.³

The proportion of subjects fulfilling the criteria of the metabolic syndrome tended to remain lower during the follow-up in both exercise groups compared with the controls. Therefore, our findings on coronary risk factors separately and on the metabolic syndrome, which is a cluster of risk factors for atherosclerosis and insulin resistance, were somewhat different. The results suggest that physical activity may prevent further clustering of metabolic risk factors also during partly unsuccessful weight maintenance.

In our study, a walking training program of moderate amount after weight reduction had a positive but minor effect on weight maintenance. Nevertheless, a walking program with a larger training volume had no independent effect on weight maintenance. Walking training in both groups seemed to prevent further clustering of metabolic risk factors. Our results support the inclusion of a moderate amount of walking in a weight maintenance program, but warns against too extensive a training regimen. The poor maintenance of both physical activity and cognitive control of eating emphasizes, however, that many issues related to long-term motivation and self-control after weight reduction remain to be solved.

Accepted for publication January 3, 2000.

This study was financially supported by the pharmaceutical companies Nycomed-Pharma AS, Oslo, Norway, and Leiras Oy, Turku, Finland; the Ministry of Education, Helsinki, Finland; and the Yrjö Jahnsson Foundation, Helsinki.

We are grateful to our exercise instructor Tuija Pokki, our nutritionist Teija Saari, MSc, and the laboratory staff at the UKK Institute for Health Promotion Research.

Corresponding author: Mikael Fogelholm, ScD, University of Helsinki, Lahti Research and Training Centre, Saimaankatu 11, FIN-15140 Lahti, Finland (e-mail: mikael.fogelholm@helsinki.fi).