Flow of patients through the Groningen Overweight and Lifestyle (GOAL) study. GP indicates general practitioner; and NP, nurse practitioner. *For example, not showing up at follow-up measurements during the screening without providing a reason or not willing to participate in further measurements like blood analyses.
ter Bogt NCW, Bemelmans WJE, Beltman FW, Broer J, Smit AJ, van der Meer K. Preventing Weight Gain by Lifestyle Intervention in a General Practice SettingThree-Year Results of a Randomized Controlled Trial. Arch Intern Med. 2011;171(4):306-313. doi:10.1001/archinternmed.2011.22
Weight regain after initial loss of weight is common, which indicates a need for lifestyle counseling aimed at preventing weight gain instead of weight loss. This study was conducted to determine whether structured lifestyle counseling by nurse practitioners (NPs) group compared with usual care by general practitioners (GP-UC) in overweight and obese patients can prevent (further) weight gain.
A randomized controlled trial in 11 general practice locations in the Netherlands of 457 patients (body mass index, 25-40 [calculated as weight in kilograms divided by height in meters squared]; mean age, 56 years; 52% female) with either hypertension or dyslipidemia or both. The NP group received lifestyle counseling with guidance of the NP using a standardized software program. The GP-UC group received usual care from their GP. Main outcome measures were changes in body weight, waist circumference, blood pressure, and fasting glucose and blood lipid levels after 3 years.
In both groups, approximately 60% of the participants achieved weight maintenance after 3 years. There was no significant difference in mean (SD) weight change and change of waist circumference between the NP and GP-UC groups (weight change: NP group, −1.2% [5.8%], and GP-UC group, −0.6% [5.6%] [P = .37]; and change of waist circumference: NP group, −0.8 [7.1] cm, and GP-UC group, 0.4 [7.2] cm [P = .11]). A significant difference occurred for mean (SD) fasting glucose levels (NP group, −0.02 [0.49] mmol/L, and GP-UC group, 0.10 [0.53] mmol/L [P = .02]) (to convert to milligrams per deciliter, divide by 0.0555) but not for lipid levels and blood pressure.
Lifestyle counseling by NPs did not lead to significantly better prevention of weight gain compared with GPs. In the majority in both groups, lifestyle counseling succeeded in preventing (further) weight gain.
trialregister.nl Identifier: NTR1365
The rising prevalence of overweight and obesity is a worldwide problem. An increased body mass index (BMI) is associated with higher mortality,1 the development of coronary vascular disease (partly independent of blood pressure [BP] and cholesterol levels2), type 2 diabetes mellitus, certain types of cancer, gastrointestinal diseases, and arthritis.3 The large impact of these diseases on shortening healthy lifespan and increasing health care costs stresses the need for strategies to tackle this problem.
Studies show that lifestyle interventions (including a nutrition and physical activity component) are needed to maintain or lose weight.4 Intensive lifestyle programs such as the Diabetes Prevention Program and the Diabetes Prevention Study showed weight losses of approximately 4 kg and 3 kg, respectively, after 3 years, accompanied by improvements in cardiovascular risk factor levels.5,6 Because weight regain after weight loss in obese persons is a common problem, a more successful approach may be to prevent weight gain and focus on weight management in those with milder degrees of overweight. Small changes in lifestyle can improve health status even without losing weight7,8 and might be easier to maintain in the long term.
The primary care setting is suitable for weight maintenance; previous studies have shown that lifestyle interventions in primary care can be effective, at least in the short term.9- 11 However, little is known about long-term effects (over several years) in this setting. Guidelines in the treatment of hypertension and dyslipidemia (often accompanied by overweight and obesity) for general practitioners (GPs) include lifestyle advice,12 but in practice, compliance with the lifestyle component of these guidelines seems limited.13 Frequently reported barriers for lifestyle counseling by GPs include lack of time, lack of patient compliance, insufficient knowledge about the subject, and lack of evidence-based interventions.14 A solution to some of these barriers may be to delegate lifestyle counseling to nurse practitioners (NPs).
The Groningen Overweight and Lifestyle (GOAL) study was conducted to compare the effects of structured lifestyle counseling by NPs with usual care by general practitioners (GP-UC) on preventing weight gain and improving health status in overweight and obese patients with either hypertension, dyslipidemia, or both.
Short-term, 1-year results of the GOAL study showed that mean weight losses in men were 2.3% in the NP group and 0.1% in the GP-UC group (P < .05), while no significant reductions were found in blood lipid levels, fasting glucose levels, and BP. In women, weight change in both groups was −1.6%. There were more individuals with weight loss (weight losers) and stable weight (stabilizers) in the NP group than in the GP-UC group (77% vs 65%) (P < .05).15
The present study reports on the long-term, 3-year results of lifestyle counseling by NPs compared with GP-UC in overweight and obese patients at relatively “low risk” for cardiovascular disease in preventing weight gain and improving health status. As a secondary objective, we investigated whether the aforementioned 1-year results were sustained after 3 years and if weight change differed within and between subgroups.
Participants were recruited (between June 2005 and February 2006) at 11 general practice locations in the northern part of the Netherlands. As previously described in detail,15 after screening and selection, 457 participants (aged 40-70 years) were enrolled within a general practice setting. Eligible participants had a BMI between 25 and 40 (calculated as weight in kilograms divided by height in meters squared) and either hypertension and/or dyslipidemia. Hypertension was defined as a mean systolic BP of 140 mm Hg or higher and/or a diastolic BP of 90 mm Hg or higher (based on 2 measurements on at least 2 different visits) or current use of BP-lowering medication, and dyslipidemia was defined as a total serum cholesterol level higher than 5.5 mmol/L (to convert to milligrams per deciliter, divide by 0.0259) or a low high-density lipoprotein cholesterol level (HDL-C) level (male, <0.9 mmol/L; female, <1.1 mmol/L) or a total cholesterol/HDL-C ratio greater than 6 and/or current use of cholesterol level–lowering medication. Exclusion criteria were diabetes mellitus, hypothyroidism, pregnancy, liver or kidney disease, current treatment for malignant disease, severely shortened life expectancy, mental illness, and addiction to alcohol or drugs. The GOAL study was approved by the Medical Ethics Review Committee of the University Medical Center Groningen, Groningen, the Netherlands, and written informed consent was given by all participants.
At the GP locations, a trained research team (not blinded for study group) performed a structured medical examination that included measurements of body weight, length, waist circumference, and BP. Body weight was measured on a digital scale with subjects wearing light clothing and no shoes, height was measured using a wall-mounted measuring tape, and waist circumference was measured at the level midway between the lowest rib and the iliac crest. Blood pressure was measured twice, and mean values were used in the analysis. The presence of cardiovascular risk factors, medication use, and family history of disease and overweight/obesity were documented. Blood samples were collected in a general practice setting after an overnight fast to analyze fasting serum lipid and glucose levels (in the same central laboratory, LabNoord in Groningen, using conventional and certified laboratory assays). Several questionnaires were completed via the Internet (as part of the software program for the lifestyle intervention) or on paper (in case of no Internet access). They contained questions on general characteristics (eg, educational level and sex of the patient) and on several issues related to body weight (eg, history of dieting). The Short Questionnaire to Assess Health-Enhancing Physical Activity (SQUASH) was used to determine physical activity.16 Metabolic syndrome was defined according to criteria from the National Cholesterol Education Program's Adult Treatment Panel III17 and Systematic Coronary Risk Evaluation (SCORE) scores to estimate 10-year risk of fatal cardiovascular disease were calculated as described by Conroy et al.18 Baseline data were available for all participants, with the following exceptions: waist circumference (n = 2), blood analyses (n = 11), complete questionnaires (n = 11), and items in questionnaire (range of missing items, 5%-11%). These missing baseline values are distributed equally among NP and GP-UC groups. The measurements were performed at baseline (between January and July 2006) after 1 and 3 years.
The number of visits in the NP and GP-UC groups was calculated via the registration system from the general practice locations. This means that both visits within the study protocol and other visits to the GP, NP, and practice assistant were included. Only visits with 1 or more components of lifestyle counseling were counted (eg, BP measurements and discussing results from blood analysis). Telephone calls and short visits were counted as 1 visit, and long visits were counted as 2 visits. The number of visits to a dietician during the study period was obtained from a separate questionnaire for the process evaluation of the study that was sent to the subjects after the 3-year measurements (50 participants did not respond to this questionnaire).
Patients were allocated by computer-generated random numbers to the NP (n = 225) or GP-UC (n = 232) groups. The lifestyle intervention consisted of 4 individual visits and 1 feedback session by telephone in the first year, in the next 2 years, 1 individual visit and 2 feedback sessions were planned each year. During these contact sessions the NP is guided by the standardized computerized software program (exclusive use for the NP group was guaranteed), which contains instructions on lifestyle counseling according to national and international guidelines19,20 and allows data entry of the measurements. The NPs followed a specially developed training program (5 sessions of 4 hours each: 4 sessions before the intervention and 1 session after 1 year) and received an individual instruction about the software program before the start of the study. The program consisted of several elements of behavioral counseling such as individual goal setting, monitoring using food diaries and pedometers, and addressing barriers for lifestyle change. The primary aim of the intervention was to prevent weight gain and if patients were motivated to lose 5% to 10% of body weight. The intervention was previously described in detail.15
The control group visited the GP after each measurement to discuss the results, and thereafter they received usual care according to GP guidelines.12
The sample-size calculation was previously described in detail.15 On the basis of previous investigations, a difference in weight loss of 2.8 kg after 1 year could be expected, resulting in the aim to include at least 145 participants in each study arm. The follow-up period in the next 2 years was meant to investigate the percentage of weight maintenance.
Differences in baseline characteristics and changes in main outcome measures after 1 and 3 years between the 2 study groups were evaluated with unpaired t tests for continuous variables and χ2 tests for categorical variables. General linear model (GLM) analysis was performed to adjust for baseline values in continuous variables. For lipid levels and BP, adjustments were made for changes in cholesterol- and BP-lowering medications, respectively. Logistic regression, with adjustments for baseline fasting glucose level, was used to examine the relation between study group and the prevalence of an impaired fasting glucose level (which is defined as a fasting glucose level >6.0 mmol/L [to convert to milligrams per deciliter, divide by 0.0555]).
Furthermore, GLM analysis was used to examine the percentage of weight loss after 1 and 3 years in subgroups of patients. Study group, sex, and each characteristic subgroup were separately entered in the model as fixed variables and age, baseline BMI, and weight change between screening and baseline as covariates. We also used GLM analysis to investigate interaction between patients' sex and study group. Regainers were defined as subjects who lost 5% of body weight or more after 1 year and returned to baseline body weight after 3 years (≥0%). Subjects were categorized into the following classes according to percentage of weight change after 1 and 3 years: successful weight losers (lost ≥5%), weight losers (weight loss from 1% to 5%), stabilizers (between >1% weight loss and 1% weight gain) and weight gainers (gain >1%). Differences in main outcome variables between these categories were tested with analysis of variance and post hoc Bonferroni test.
Results are presented with exclusion of dropouts and missing values and were adjusted for baseline values. Thereafter, all analyses were also performed following the intention-to-treat principle by BOCF (baseline observation carried forward) for dropouts. Usually, BOCF indicates that there is no weight change, so in our study this would mean that dropouts were characterized as successful because they did not gain 1% or more of their body weight. But this might be an overestimation of the percentages of the participants who achieved weight maintenance. Therefore we also performed analyses in which all dropouts were considered as not successful, which is probably an underestimation of the success rate. Percentages of participants who achieved weight maintenance are presented as a range of both methods.
All analyses were performed in 2009 using SPSS/PC statistical program version 16.0 for Windows (SPSS Inc, Chicago, Illinois). P < .05 was considered statistically significant.
The dropout percentage was 24% for the NP and 20% for the GP-UC group (not significant, P = .28) (Figure). Participants who dropped out had a higher diastolic BP (89 vs 86 mm Hg; P = .003) and more often had a BMI of 30 or greater (reasons of dropout did not significantly differ between participants with a BMI above or under 30); there were no differences in other characteristics between dropouts and participants who attended the follow-up measurement after 3 years. Table 1 gives the baseline characteristics for both groups. Participants in the GP-UC group more often had greater than 3 recent dieting attempts than in the NP group and more frequently fulfilled the norm on physical activity (P < .05 for both).
After 3 years, no differences in mean (SD) weight change were present between the NP and GP-UC groups (NP, −1.2% [5.8%]; GP-UC: −0.6% [5.6%]; P = .37). Approximately 60% of the participants in both groups were weight losers or stabilizers after 3 years (Table 2).
In the NP group, a positive effect was found on mean (SD) fasting glucose level at the 3-year follow-up compared with the GP-UC group (−0.02 [0.49]) mmol/L vs 0.10 (0.53) mmol/L; P = .02). After 3 years the prevalence of impaired fasting glucose level was 6% (n = 10) in the NP group vs 12% (n = 21) in the GP-UC group (P = .17) (Table 2). In persons with a BMI of 30 or greater, the prevalence was lower in the NP than in the GP-UC group (7% [n = 4] vs 22% [n = 14]) but was not significant after adjustment for baseline values (P = .14) (data not shown). No significant differences between the NP and GP-UC groups occurred for serum lipid levels and BP at the 3-year follow-up (Table 2).
Contrary to the results after 3 years, the percentage of mean (SD) weight loss after 1 year differed between the NP and GP-UC groups (NP group, −2.2% [7.0%]; GP-UC group, −0.7% [4.6%]; P = .002), and there were more weight losers and stabilizers after 1 year in the NP group than in the GP-UC group (80% vs 64%; P = .001) (Table 3). The percentage of regainers in the NP group was comparable to the GP-UC group (14% and 16%, respectively) (Table 2).
There were no differences in weight change after 3 years between the NP and GP-UC groups in subgroups of patients' characteristics at baseline (Table 3). Interaction between sex and study group was absent. Within the NP group, participants with 3 or fewer attempts to lose weight during the last 5 years lost more weight after 3 years than participants with 4 or more attempts (P < .05). These participants gained 1.4% (95% confidence interval, −0.9 to 3.7) of their baseline weight after 3 years.
Successful weight losers achieved the most favorable results and weight gainers the least favorable results after 3 years on physiological outcome variables except for systolic BP (Table 4). After 3 years, stabilizers and weight gainers (n = 192) achieved significantly better results for mean (SD) fasting glucose level (0.10 [0.50] mmol/L) than regainers (0.57 [0.60] mmol/L) (P = .008) but not for lipid levels and BP (data not shown).
Intention-to-treat analysis did not substantially alter the results after 3 years. For example, mean (SD) weight loss after 3 years in the NP group was −0.8% (5.0%) and −0.5% (5.0%) in the GP-UC group (P = .45). Change in mean (SD) low-density lipoprotein cholesterol (LDL-C) level did not differ between the NP and GP-UC groups (NP group, 0.15 [0.71] mmol/L; GP-UC group, 0.04 [0.77] mmol/L; P = .11). Change in mean (SD) fasting glucose level after 3 years differed significantly between the NP and GP-UC groups (−0.01 [0.43] mmol/L vs 0.08 [0.47] mmol/L, respectively; P = .03). The percentages of participants who achieved weight maintenance varied from 47% to 71% in the NP group and from 51% to 71% in the GP-UC group (depending on whether dropouts are considered successful or not).
The design of our study was different from several other published weight intervention studies because we focused on weight maintenance in persons with a relatively low mean BMI (almost 30) and because we had a considerably longer follow-up than in most other studies, within a “realistic” primary care setting. The relevance of prolonged follow-up is reflected by the differences between the 1-year and the somewhat disappointing 3-year results on weight maintenance. After 1 year, 80% of the participants in the NP group indeed achieved weight maintenance vs 64% in the GP-UC group. However, after 3 years, differences between both groups had disappeared: with 60% success in weight maintenance of the participants in both groups. Changes in fasting glucose levels differed in favor of the NP group, especially among obese persons the prevalence of an impaired fasting glucose level differed considerably after 3 years (7% vs 22%), but owing to lack of power in this subgroup analysis, this result was not significant when adjusted for baseline values (P = .14).
In comparison with other studies with prolonged follow-up, Jeffery and French21 described a weight gain of approximately 1.5 kg after 3 years for treatment and control groups, but they used a low-intensity intervention mostly conducted by mail. Another randomized controlled trial with a longer follow-up of 54 months was reported by Simkin-Silverman et al,22 but this was done in postmenopausal women with normal weight and cardiovascular risk and showed that weight maintenance is possible with a lifestyle intervention. It is notable that we found a small difference in the fasting glucose level, while in the Diabetes Prevention Study, differences between intervention and control groups were absent for fasting glucose levels, although significant differences in weight change did occur.6 Other studies have shown that changes in lifestyle without losing weight can improve health status.7,8
Most of the participants in both groups achieved weight maintenance. Several factors may be responsible for the long-term lack of difference that we expected between the NP and GP-UC groups. Patients in the GP-UC group may be more adherent to advice given by the GP because of the study circumstances. In line with the ethical committee demands, all patients were fully informed about the study purpose, and hence they knew beforehand that body weight was assessed as well as adherence to lifestyle advice. This in itself may, through some kind of Hawthorne effect, lead to modified behavior so that all patients were more adherent than they might have been under other circumstances, and this may have diluted any differences between the groups. Moreover, both groups in this trial comprised volunteers and were thus a selection of motivated patients.
The attention on health (and body weight) during the measurements in combination with abundant countrywide campaigns for a healthy lifestyle held during the course of the study may, besides the visits to the GP, also have been responsible for lifestyle changes in the control group. In comparison with the Dutch population, where an average increase in BMI of 0.05 per year (between 1981 and 2004) was described by Gast et al,23 we found a decrease of −0.4 in the NP group and −0.2 in the GP-UC group. Thus, we can consider that the majority in the NP and GP-UC groups succeeded in preventing (further) weight gain.
Besides the limitations like baseline differences between NP and GP-UC groups and randomization at a patient level instead of at a practice level, as previously described in detail,15 another limitation of the GOAL study needs to be discussed. The visits to the NP after the first year occurred at a low frequency and may not be sufficient to sustain weight loss. Overall, in both groups, achieving weight maintenance was not influenced by the mean number of visits (Table 3). In the Diabetes Prevention Study, 4 face-to-face visits each year were scheduled after the first year to achieve sustained weight loss after 3 years.6 Bogers et al24 also described that higher intervention costs (indicative for the intensity of an intervention) are associated with greater weight loss. Although this association with weight loss was determined after 1 year, it is plausible that it will also apply for long-term weight loss.
In intention-to-treat analysis regarding weight change, we chose a conservative way to deal with the dropouts by carrying the baseline observation forward. This means that we assumed that dropouts during the intervention lost no weight or regained all the weight that might be lost in the first year of the intervention and thereby possibly underestimated the weight loss of the dropouts in both groups.
The strengths of our study are the large study population with an equal amount of male and female participants, a relatively low dropout after 3 years, the prolonged follow-up, and the use of an intervention that is feasible in a primary care setting. The software program can easily be used at other locations, and the intervention is not time intensive and expensive. More research is planned to evaluate the process of the GOAL intervention, which is useful for further implementation.
Analyses in subgroups showed that within the NP group participants with 3 or fewer recent attempts to lose weight had a lower weight after 3 years compared with participants with more than 3 attempts. The latter participants' mean weight gain was 1.4% (95% confidence interval, −0.9% to 3.7%). This means that our intervention is not suitable for experienced dieters.
Regainers achieved unfavorable results on fasting glucose levels compared with stabilizers and weight gainers, which is in line with other negative health effects of weight cycling that were described.25- 27 No clear results have been reported on the relation between repeated weight losses and mortality and the underlying mechanisms.28- 31
We can conclude that preventing weight gain by NPs did not lead to significantly better results than by GPs. More follow-up sessions in the NP group may lead to a higher percentage of maintenance of the weight that was lost after 1 year.
Correspondence: Nancy C. W. ter Bogt, MSc, Sector F: Department of General Practice, University Medical Center Groningen, A. Deusinglaan 1, 9713 AV Groningen, the Netherlands (firstname.lastname@example.org).
Accepted for Publication: July 20, 2010.
Author Contributions: Dr van der Meer had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Bemelmans, Beltman, Broer, Smit, and van der Meer. Acquisition of data: ter Bogt, Beltman, and Broer. Analysis and interpretation of data: ter Bogt, Bemelmans, Beltman, Smit, and van der Meer. Drafting of the manuscript: ter Bogt and Smit. Critical revision of the manuscript for important intellectual content: Bemelmans, Beltman, Broer, Smit, and van der Meer. Obtained funding: Bemelmans, Beltman, and Broer. Administrative, technical, and material support: ter Bogt, Beltman, and Broer. Study supervision: Bemelmans, Beltman, Smit, and van der Meer.
Financial Disclosure: None reported.
Funding/Support: The GOAL study was financially supported by the Netherlands Organization for Health Research and Development (Zon-Mw, project No. 6200.0016) and Foundation Fund “De Gavere.”
Role of the Sponsors: The funders had no role in the design and conduct of the study; collection, management, analyses, and interpretation of the data; and preparation, review, or approval of the manuscript.
Additional Contributions: We are grateful to the participating general practices and their patients for their enthusiasm and cooperation. Thanks to all the students of the research team for performing measurements and data collection. We also thank the members of the Hypertension Service Groningen for generating the GOAL-study. Neither received compensation.