Obesity is a common health problem that requires a long-term care approach. We systematically reviewed long-term (≥2 y) studies investigating dietary/lifestyle, pharmacologic, and surgical weight loss methods to assess (1) weight loss efficacy, defined by absolute weight loss and the proportion of subjects with ≥5% weight loss, (2) effects of weight loss on cardiovascular risk factors, and (3) applicability of findings from studies to everyday clinical practice.
The MEDLINE, HealthSTAR, and the Cochrane Controlled Trials databases were searched for studies investigating the long-term efficacy of weight loss methods in overweight and obese adults. Data were extracted for (i) weight loss after 1 y (pharmacologic studies only), 2 y, 3 y, and 4 y, (ii) proportion of subjects with ≥5% weight loss at the end of follow-up, and (iii) changes (end-of follow-up minus baseline values) in blood lipids, fasting blood glucose, and systolic and diastolic blood pressure.
Dietary/lifestyle therapy provides <5 kg weight loss after 2–4 y, pharmacologic therapy provides 5–10 kg weight loss after 1–2 y, and surgical therapy provides 25–75 kg weight loss after 2–4 y. Weight loss of ≥5% baseline weight is not consistently associated with improvements in cardiovascular risk factors and these benefits appear to be intervention specific and occur mainly in people with concomitant cardiovascular risk factors. Weight loss studies have methodologic limitations that restrict the applicability of findings to unselected obese people assessed in everyday clinical practice. These limitations include an inadequate study duration, large proportions of subjects lost to follow-up, a lack of an appropriate usual care group, and a lack of reporting of outcomes in high-risk subgroups.
Dietary/lifestyle and pharmacologic weight loss interventions provide modest weight loss, and may improve markers of cardiovascular risk factors although these benefits occur mainly in patients with cardiovascular risks. Studies investigating weight loss have methodologic limitations that restrict the applicability of findings to obese patients assessed in clinical practice.
Obesity is a major health care priority of the 21st century.1, 2, 3 Obesity is associated with an increased risk for cardiovascular disease, type 2 diabetes, hypertension, dyslipidemia, and other diseases.4, 5, 6, 7 In 2000, 30% of adults in the United States were obese, with a body mass index (BMI) ≥30 kg/m2, and 34% were overweight, with a BMI 25–29.9 kg/m2, and at risk for obesity.8 The prevalence of obesity in the United States and other industrialized countries has increased by 50–60% since 1980.9, 10, 11, 12
Numerous short-term studies, usually of 6 months duration, have investigated weight loss. However, obesity is a chronic condition, like diabetes or hypertension.13, 14, 15, 16, 17, 18 Life-long management is required and long-term studies of efficacy are most relevant. Also, weight regain will be undetected in short-term studies because it occurs slowly in over 80% of people who lose weight.2, 19, 20 Clinically important weight loss, defined by obesity experts as loss of 5–10% of baseline weight,21, 22, 23, 24 may improve lipid, glucose, and blood pressure levels,25, 26, 27 with potential reductions in cardiovascular disease.28, 29, 30 However, it is unclear if weight loss improves risk factors in all obese persons or only in high-risk groups. Finally, it is important to determine if weight loss studies are applicable to the everyday clinical management of obese patients.
We, therefore, systematically reviewed long-term studies investigating dietary/lifestyle, pharmacologic, and surgical methods of weight loss to assess (1) weight loss efficacy, defined by absolute weight loss and the proportion of subjects with ≥5% weight loss, (2) effects of weight loss on cardiovascular risk factors, and (3) applicability of findings from studies to everyday clinical practice.
The MEDLINE (1966–September 2003), HealthSTAR (1975–September 2003), and Cochrane Controlled Trials Register (1990–September 2003) databases were searched for weight loss studies in adults, aged 18–65 y. Search strategies used the key words obesity, overweight, body mass index, treatment, weight reduction, randomized controlled trial, and cohort study. English and non-English articles were searched to avoid language-related selection bias,31 but only articles with an English abstract were reviewed. This search was supplemented by a review of relevant bibliographies and consultation with experts to identify overlooked articles.32
Studies were identified using prespecified methods.33, 34 Eligible studies satisfied three criteria: (1) the study population consisted of overweight or obese adults with a BMI ≥25 kg/m2; (2) the study investigated a prespecified dietary/lifestyle, pharmacologic, or surgical weight loss method; and (3) the study was a clinical trial (ie randomized controlled or nonrandomized). Eligible studies with one or more of these characteristics were excluded: (1) nonconsecutive (or selected) subjects were studied, such as those required to attain weight loss targets prior to enrollment; (2) <100 subjects were studied, since skewed results that misrepresent a treatment effect may occur in small studies;35 (3) the intervention is not approved for clinical use (eg fenfluramine)36 or is approved only for short-term use (eg benzphetamine);37 and (4) the duration of subject follow-up was <2 y. (A 1 y of follow-up was allowed for pharmacologic studies because only four studies were identified with a follow-up of ≥2 y.) Eligible studies were reviewed independently by two authors (JD, CM) to assess inclusion suitability and data extraction accuracy. Interobserver agreement was assessed with the kappa (κ) statistic,38 and disagreements were resolved by consensus.
Study classification and characteristics of weight loss methods
Studies were classified according to the principal weight loss intervention. (1) Dietary/lifestyle: a regimen consisting of a very-low-calorie diet (<1100 kcal/day), consisting typically of liquid meal replacements, or a low-calorie diet (1200–1500 kcal/day), consisting typically of a nutritionally balanced hypocaloric diet or prepackaged foods. Dietary/lifestyle therapy may be combined with a behavior counselling/lifestyle modification and/or an exercise program. (2) Pharmacologic: drug therapy with orlistat, a gastrointestinal lipase inhibitor that restricts fat absorption, or sibutramine, a serotonin–norepinephrine re-uptake inhibitor that enhances satiety and increases basal energy expenditure. Drug therapy is combined, typically, with dietary therapy that is usually a low-calorie diet. (3) Surgical: surgical therapy consisting of a restrictive procedure that induces early satiety (eg vertical banded gastroplasty, gastric banding) or a diversionary procedure that decreases caloric absorption (eg gastric bypass). Surgical therapy is usually combined with dietary therapy and/or a behavior counselling/lifestyle modification program and, typically, is limited to people with class II obesity (BMI ≥35 kg/m2) and severe comorbidity or class III obesity (BMI ≥40 kg/m2).
For each weight loss method in each study, we extracted data (whenever possible) in three domains: (1) weight loss efficacy: body weight (kg) at baseline, after 1 y (pharmacologic studies only), 2, 3, and 4 y; and weight loss of ≥5% at the end of the follow-up. Weight loss data were extracted separately for the ‘last-observation-carried-forward’ method of analysis, which documents weight loss for all subjects regardless of when they leave the study, and for the ‘study-completers’ method of analysis, which documents weight loss only for subjects who completed follow-up. (2) Effects on cardiovascular risk factors: we assessed the effect of weight loss on change in cardiovascular risk factor levels, from baseline to the end of follow-up, for five risk factors that are considered clinically important:39, 40, 41 (i) low-density lipoprotein (LDL) cholesterol; (ii) high-density lipoprotein (HDL) cholesterol; (iii) triglycerides; (iv) fasting blood glucose; (v) systolic and diastolic blood pressure. (3) Applicability to clinical practice: we assessed four study design criteria that we considered relevant to the applicability of findings to clinical practice: (i) duration of study; (ii) proportion of subjects lost to follow-up; (iii) appropriate control or ‘usual care’ group; and (iv) weight loss effects on cardiovascular risk factors in high-risk groups. We also assessed study quality using a prespecified grading system outlined in Appendix A1.34
Meta-analysis was used to pool the results of studies investigating the efficacy of orlistat and sibutramine compared with placebo, measured by the proportion of subjects with ≥5% weight loss after 1 or 2 y of treatment. For fixed and random effects models, a weighted summary estimate of treatment effect was obtained and was expressed as an odds ratio (OR) and 95% confidence interval (CI). The χ2 test was used to assess heterogeneity across studies, and in cases of significant heterogeneity (P<0.10) the results were based on the random effects model.42 Meta-analysis was not appropriate for dietary and surgical studies, because these studies differed in their interventions, nor for metformin therapy studies, because only two studies were available. A descriptive analysis was used to summarize these studies. Meta-analysis was not feasible to assess the effects of weight loss on markers of cardiovascular risk because there was no consistency in the markers assessed across studies nor consistent availability of measures of variation (ie standard deviation or standard error) for point estimates across studies. The applicability of findings to clinical practice was based on a subjective assessment of studies that satisfied the prespecified criteria.
The study selection process outlined in Figure 1 identified 44 studies.43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89 There was fair or good agreement for study inclusion suitability (κ=0.67) and data extraction (κ=0.71). Study quality was rated as fair or good in dietary/lifestyle and pharmacologic studies, and fair or poor in surgical studies (Tables 1, 2 and 3).
In 16 studies that involved 5698 subjects, the mean age and weight of subjects across studies varied from 40 to 59 y and from 78 to 116 kg (Table 1). In 13 studies, weight loss was assessed during a 3- to 18-month weight loss induction phase, when subjects received an intervention, followed by a 12- to 36-month weight loss maintenance phase after the intervention. In three studies, weight loss was assessed only during a 24- to 48-month induction phase.53, 56, 57
Weight loss efficacy
Weight loss was usually <5 kg after 2–3 y (mean (±s.d.): 3.5±2.4 kg; range: 0.9–10.0 kg) and after 4–7 y (mean: 3.6±2.6 kg; range: 1.8–10.0 kg) based on a completers method of analysis. There was no subjective difference in efficacy for studies with only an induction phase (weight loss: 1.5–3.5 kg) compared to studies with an induction and maintenance phase (weight loss: 0.9–10.4 kg). Although no study reported on ≥5% weight loss, it is likely that only a small proportion of subjects attained this target given the baseline weight (∼100 kg) and weight loss (<5 kg) of subjects across studies.
Effects on cardiovascular risk factors
There was a trend across studies of decreased blood pressure with dietary/lifestyle therapy, but it was statistically significant in some47, 54, 55, 56 but not other48, 51, 52, 57 studies. In studies of subjects with impaired glucose tolerance or type 2 diabetes, <5 kg weight loss conferred significant reductions in systolic (5–7 mmHg) and diastolic (5–6 mmHg) blood pressure,47, 56 whereas in studies of subjects with hypertension, the effects on blood pressure control were inconsistent.51, 52, 57 In one study, more subjects with diet counselling remained off antihypertensive drug therapy than subjects without counselling (39 vs 5%; P<0.001) although blood pressure control was better in subjects receiving drug therapy.57 In three studies of subjects with impaired glucose tolerance or type 2 diabetes who received dietary/lifestyle and exercise counselling or usual care, counselling conferred a 58–63% lower risk of type 2 diabetes.47, 56, 77
Applicability to clinical practice
Six studies had a follow-up of at least 4 y. Nine studies had high proportion (31–64%) of subjects lost to follow-up. This may affect clinical applicability of findings because in all but two studies,50, 77 outcomes were reported based on a ‘study-completers’ analysis, which may overestimate weight loss because study completers may have greater weight loss than noncompleters. Six studies had a ‘usual care’ control group.47, 51, 52, 53, 56, 59, 77 No study reported on weight loss and its effects of cardiovascular risks in high-risk groups, unless the study was limited to high-risk subjects.47, 51, 52, 56, 57, 77
In 19 studies that involved 9953 subjects, the mean age and weight of subjects across studies varied from 40 to 58 y and from 87 to 102 kg (Table 2). In all studies, subjects received a dietary/lifestyle regimen and either drug therapy or placebo, weight loss was assessed during drug therapy, and weight loss maintenance was not assessed after drug therapy was stopped.
Weight loss efficacy
Weight loss was usually 5–10 kg after 1 y (mean: 6.1±2.0 kg; range: 3.9–10.3) and after 2 y (mean: 7.2±1.6 kg; range: 4.9–8.9) based on a last-observation-carried-forward method of analysis. Weight loss of ≥5% occurred in 40–60% of subjects, and those who received pharmacologic therapy had a three-fold greater odds of attaining such weight loss than subjects who received diet-only therapy (OR=2.94; 95% CI: 2.47–3.50) (Figure 2). Orlistat therapy increased the odds of attaining ≥5% weight loss compared with diet-only therapy after 1 y (OR=2.54; 95% CI: 2.17–2.90) and 2 y (OR=4.55; 95% CI: 1.99–10.4). Sibutramine therapy increased the odds of attaining ≥5% weight loss compared with diet-only therapy after 1 y (OR=5.03; 95% CI: 2.27–11.1) and 2 y (OR=4.44; 95% CI: 1.87–10.5). Metformin therapy was associated with weight loss of <3 kg, although its primary effect was to improve glycemic control. In one study that compared metformin therapy with a lifestyle modification program, weight loss was greater with lifestyle modification (5.6 vs 2.1 kg; P<0.01).77
Effects on cardiovascular risk factors
Pharmacologic therapy improved lipid levels, and glycemic and blood pressure control, but these effects were drug specific and greatest in subjects with elevated baseline levels. Orlistat therapy was associated with modest but significantly greater reductions in LDL levels (0.11–0.38 mmol/l) than diet-only therapy, but there were no significant effects on HDL and triglyceride levels except in one study.72 No significant effect on blood pressure compared with dietary/lifestyle-only therapy occurred, except in two studies, one of which involved subjects with poorly controlled hypertension.62, 71 Orlistat had inconsistent effects on glycemic control: modest but significantly greater reductions in fasting blood glucose (0.1–1.7 mmol/l) than diet-only therapy in six studies,60, 61, 62, 63, 67, 68 but no difference in two studies.65, 66 The greatest improvements in glycemic control occurred in subjects with type 2 diabetes.67, 68 There were no significant effects of sibutramine therapy compared to diet-only therapy on lipids and glycemic control, but sibutramine was associated with higher systolic (1–4 mmHg) and diastolic (2–4 mmHg) blood pressure,72, 73, 75, 76 including subjects with treated hypertension.73, 75 Metformin had no effect on lipids or glycemic or blood pressure control in one study,78 but in another study of subjects with impaired glucose tolerance,77 metformin therapy conferred a 31% lower risk for type 2 diabetes.
Applicability to clinical practice
All studies except one77 had a follow-up of 2 y or less. The 2 -y studies had a high proportion of subjects lost to follow-up (30–57%). This may affect clinical applicability of findings because outcomes in all studies were reported based on the ‘last-observation-carried-forward’ method, which may overestimate weight loss because outcomes before subjects leave the study are documented, and such subjects may be destined to regain lost weight. In all studies, there was an appropriate ‘usual care’ group who received placebo. No study reported on weight loss and its effects of cardiovascular risk factors in high-risk groups, unless the study was limited to high-risk subjects.67, 68, 70, 71, 73, 75, 76, 77
In nine studies that involved 3622 subjects, the mean age and weight of subjects across studies varied from 34 to 49 y and from 110 to 142 kg (Table 3). All subjects received a dietary regimen, although this was usually not specified.
Weight loss efficacy
Weight loss was 28–76 kg after 2 y,79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89 17–73 kg after 3 y,79, 80, 81, 86, 87, 88, 89 and 20 kg after 8 y in one study82, 83, 84 based on a completers method of analysis. No study reported the proportion of subjects with ≥5% weight loss, although this was probably high given the absolute weight loss with surgical therapy.
Effects on cardiovascular risk factors
Only two studies assessed this,82, 83, 84, 88 thereby precluding meaningful commentary. In one study that compared surgical and nonsurgical weight loss after 8 y, the risk of type 2 diabetes was 84% less with surgery (OR=0.16; 95% CI: 0.07–0.36) but there was no effect on the risk of hypertension (OR=1.01; 95% CI: 0.61–1.67).82, 83, 84
Applicability to clinical practice
Three studies had a follow-up of at least 4 y.80, 82, 83, 84, 89 The proportion of subjects lost to follow-up varied widely (5–42%), but was reported only in three studies. One study had a control group of ‘usual care’. No study reported on weight loss in high-risk groups.
There are three conclusions from this review: (1) in terms of weight loss efficacy, dietary/lifestyle therapy provides <5 kg weight loss after 2–4 y, pharmacologic therapy provides 5–10 kg weight loss after 1–2 y, and surgical therapy provides 25–75 kg weight loss after 2–4 y; (2) weight loss of ≥5% is not consistently associated with improvements in cardiovascular risk factors, and when improvements occur they are mainly in high-risk groups and appear to be intervention specific; and (3) weight loss studies have methodologic limitations that restrict their application to everyday clinical practice.
Our finding that dietary/lifestyle therapy is associated with only modest (<5 kg) weight loss, although surprising, was corroborated in the dietary-only arms of pharmacologic therapy studies, in which weight loss was 1.7–4.9 kg. Although greater weight loss occurs when pharmacologic agent is added to dietary/lifestyle therapy, this may require continuous drug therapy, as weight regain can occur after treatment is stopped.90 However, orlistat and sibutramine are not approved for longer than 1 and 2 y, respectively.91 The large weight loss from surgery may depend on ongoing dietary/lifestyle interventions,92, 93 and should be assessed against a 0.5–1% postoperative mortality risk and an increased risk of wound dehiscence, venous thromboembolism, and cardiorespiratory insufficiency.94 Newer laparoscopic techniques, however, are associated with fewer postoperative complications.94, 95
The impact of weight loss on cardiovascular risk factors is greatest in subjects with impaired glucose tolerance, type 2 diabetes, or hypertension, because changes in risk factors are more likely in subjects with abnormal baseline levels, and this is consistent with findings from observational studies that intentional weight loss is associated with increased longevity but only in people with pre-existing disease.96, 97 We found, however, that these improvements may be intervention specific. Orlistat therapy was associated with a 5–10% decrease in LDL levels, which is comparable to a cholesterol-lowering diet.98, 99 Some dietary/lifestyle interventions improved glycemic control and lowered blood pressure similar to antihypertensive drug therapy,100 but without affecting lipids. Overall, our findings suggest that ≥5% weight loss, irrespective of patient characteristics and intervention, does not uniformly improve cardiovascular risk factors. However, there is a lack of data to correlate weight loss with effect on markers of cardiovascular risk, as there may be weight loss thresholds that, depending on the study population, are associated with more consistent health benefits.
A key issue is the restricted clinical applicability of the study findings. Few studies have longer than 3 y of follow-up, whereas studies of hypertension or diabetes management typically have 4–8 y of follow-up to assess potential improvements in cardiovascular outcomes.100, 101 Losses to follow-up were typically 30–60%, whereas <20% loss to follow-up, a minimum requirement for clinical trials,34, 102 was satisfied by only six studies. With such high losses to follow-up, the ‘study-completers’ and ‘last-observation-carried-forward’ methods of analysis may overestimate weight loss efficacy and improvements in cardiovascular risk factors because they do not document weight regain that occurs after subjects leave a study. Only seven of 25 dietary/lifestyle and surgical therapy studies included a ‘usual care’ control group, which limits assessment of weight loss efficacy and impact on cardiovascular risk factors that might occur in clinical practice. Studies did not report the effects of ≥5% weight loss on markers of cardiovascular risk factors in subjects with or without cardiovascular risk factors. Such information would assist in deciding about which patients would benefit most from weight loss. Furthermore, prior knowledge about the clinical significance of such weight loss in improving quality of life and reducing cardiovascular outcomes would strengthen the rationale for weight loss in high-risk groups. Overall, there is a need for standardized reporting criteria in weight loss studies to improve applicability of results to everyday clinical practice, and a proposed framework is outlined in Table 4.
Our review can be differentiated from other reviews of weight loss, which included studies with a short follow-up duration, did not specify the study selection process, or did not assess applicability to clinical practice.7, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116 Our review has some limitations. We did not assess supplementary weight loss methods, such as behavioral or exercise programs, when used as stand-alone interventions. In general, these interventions provide modest weight loss (<5 kg), although exercise is important for weight loss maintenance and may reduce cardiovascular risk factors independent of weight loss.117, 118 We also did not assess the mechanisms, benefits, and risks of specific weight loss methods, which have been addressed elsewhere.7, 103, 114, 115, 116
Our findings have potential clinical implications. First, in obese people with cardiovascular risk factors, weight loss can provide some health benefits. However, the effects of weight loss on cardiovascular risk factors are modest and inconsistent, and are predicated on sustained, long-term, weight loss. These limitations suggest the need for additional interventions (eg lipid-lowering and antihypertensive drug therapy) to attain long-term, clinically significant improvements in risk factors. Second, in obese people without cardiovascular risk factors, maintenance of a stable weight is reasonable given the limited efficacy of weight loss interventions. Third, because weight loss is difficult to achieve and maintain, the prevention of obesity, rather than its treatment, may be a more worthwhile clinical focus.
To summarize, dietary/lifestyle therapy provides <5 kg weight loss after 2–4 y, pharmacologic therapy provides 5–10 kg weight loss after 1–2 y, and surgical therapy provides 25–75 kg weight loss after 2–4 y. Weight loss of ≥5% baseline weight improves lipid levels and glycemic and blood pressure control, but these benefits occur mainly in people with cardiovascular risk factors and appear to be intervention specific. Most striking, perhaps, are methodologic limitations in weight loss studies that restrict the applicability of findings to everyday clinical practice. Additional studies are needed that use standardized criteria to assess long-term weight loss efficacy and clinical significance.
We thank Drs Arya Sharma, Hertzel Gerstein, and Anna Sawka for their helpful reviews of the manuscript.
Appendix A1 Criteria for grading the internal validity of individual studies
(Adapted from the Process Manual for Producing and Disseminating CTF Reviews, Canadian Task Force on Preventive Health Care (May 2002 draft), and from the United States Preventive Services Task Force Methods Manual (August 2000 draft).
Design-specific criteria and quality category definitions
Presented below are a set of minimal criteria for each study design and then a general definition of three categories: ‘good’, ‘fair’, and ‘poor’ based on those criteria. These specifications are not meant to be rigid rules, but rather are intended to be general guidelines, and individual exceptions, when explicitly explained and justified, can be made. In general,
a ‘good’ study is one that meets all criteria well
a ‘fair’ study is one that does not meet (or it is not clear that it meets) at least one criterion but has no known ‘fatal flaw’
Randomized controlled trials and cohort studies
Initial assembly of comparable groups.For randomized controlled trials (RCTs): adequate randomization, including first concealment and whether potential confounders were distributed equally among groups
For cohort studies: consideration of potential confounders with either restriction or measurement for adjustment in the analysis; consideration of inception cohorts
Maintenance of comparable groups (includes attrition, crossovers, adherence, contamination)
Important differential loss to follow-up or overall high loss to follow-up
Measurements: equal, reliable, and valid (includes masking of outcome assessment)
Clear definition of interventions
All important outcomes considered
Analysis: adjustment for potential confounders for cohort studies, or intention to treat analysis for RCTs.
Definition of ratings based on above criteria
Good: Meets all criteria. Comparable groups are assembled initially and maintained throughout the study (follow-up at least 80%); reliable and valid measurement instruments are used and applied equally to the groups; interventions are spelled out clearly; all important outcomes are considered; and appropriate attention to confounders in analysis. In addition, for RCTs, intention to treat analysis is used.
Fair: Studies will be graded ‘fair’ if any or all of the following problems occur, without the fatal flaws noted in the ‘poor’ category below: generally, comparable groups are assembled initially but question remains whether some (although not major) differences occurred with follow-up; measurement instruments are acceptable (although not the best) and generally applied equally; some but not all important outcomes are considered; and some but not all potential confounders are accounted for. Intention to treat analysis is carried out for RCTs.
Poor: Studies will be graded ‘poor’ if any of the following fatal flaws exists: groups assembled initially are not close to being comparable or maintained throughout the study; unreliable or invalid measurement instruments are used or not applied at all equally among groups (including not masking outcome assessment); and key confounders are given little or no attention. For RCTs, intention to treat analysis is lacking.
About this article
Impact of weight loss-associated changes in detailed body composition as assessed by whole-body MRI on plasma insulin levels and homeostatis model assessment index
European Journal of Clinical Nutrition (2017)