A systematic review of reviews: exploring the relationship between obesity, weight loss and health-related quality of life.

This is the first systematic review of reviews to assess the effect of obesity and weight loss on health-related quality of life (HRQoL). We identified 12 meta-analyses/systematic reviews published between January 2001 and July 2016. They addressed the following themes: (i) the relationship between weight/body mass index and HRQoL (baseline/pre-intervention; n = 2). (ii) HRQoL after weight loss (varied interventions and/or study design; n = 2). (iii) HRQoL after weight loss (randomized controlled trials only; n = 2). (iv) HRQoL after bariatric surgery (n = 6). We found that in all populations, obesity was associated with significantly lower generic and obesity-specific HRQoL. The relationship between weight loss and improved HRQoL was consistently demonstrated after bariatric surgery, perhaps due to a greater than average weight loss compared with other treatments. Improved HRQoL was evident after non-surgical weight loss, but was not consistently demonstrated, even in randomized controlled trials. This inconsistency may be attributed to variation in quality of reporting, assessment measures, study populations and weight-loss interventions. We recommend longer-term studies, using both generic and obesity-specific measures, which go beyond HRQoL in isolation to exploring mediators of HRQoL changes and interactions with other variables, such as comorbidities, fitness level and body image.

Introduction

People with obesity have an increased risk of a multitude of diseases and early mortality 1, 2, 3, 4, 5. Beyond health risks, obesity has also been shown to negatively impact quality of life (QoL) 6, defined as an individual's own assessment of well‐being, often with reference to physical and mental health status, social relationships and environmental and economic factors 7, 8, 9. When the focus in clinical and health research is on the quantification of QoL related to health status, it is referred to as health‐related QoL (HRQoL) 7, 8, 10. This term is most commonly understood to refer to a multidimensional measurement of the individual's perception of the impact of illness and its treatment 10, 11. HRQoL captures, at a minimum, physical, psychological and social functioning 10. The various measures for assessing different aspects of an individual's HRQoL can be grouped into two categories: generic measures and disease‐specific measures 7, 12. Generic measures assess broad aspects of HRQoL, while disease‐specific measures are designed to assess HRQoL in relation to a specific medical condition or clinical population.

Because studies on obesity and/or weight loss increasingly include measures of HRQoL, a large number of review articles have been written on this topic. We considered that a systematic review of review articles, comprising a comprehensive examination of the current state of the field, is needed to evaluate the impact of obesity and weight loss on HRQoL. According to Smith et al., one goal of a systematic review of reviews is to identify and appraise all published reviews within an area of interest in order to summarize and compare conclusions, as well as discuss the strength of these conclusions 13. The aim of this review is to synthesize the information found in obesity and/or weight‐loss reviews that assess HRQoL and to evaluate the impact of obesity and weight reduction on HRQoL.

Methods

We conducted a systematic review of reviews in accordance with the Preferred Reporting Items for Systematic Review and Meta‐Analysis (PRISMA) guidelines (http://www.prisma‐statement.org/). The PubMed and Embase databases were used to identify review articles that have evaluated the evidence of the impact of obesity and weight management on HRQoL (Fig. 1). The search terms applied to the PubMed and Embase databases are detailed in Fig. 2. A search string was defined and limits were applied (Fig. 2). The search was restricted to January 2001 to July 2016. The selection of review articles to include was performed by three independent reviewers (RK; JRA; AXON) and articles were excluded if they met specific predefined exclusion criteria (Fig. 2).

Figure 1

Preferred Reporting Items for Systematic Review and Meta‐Analysis (PRISMA) flow chart. NA, not applicable. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 2

Identification and selection of published review articles on obesity and/or weight management and quality of life from January 2001 to July 2016. QoL, quality of life. [Colour figure can be viewed at wileyonlinelibrary.com]

Results

A total of 540 review articles were retrieved from PubMed and Embase, of which 12 were selected for inclusion in this systematic review of reviews. Table 1 provides a detailed overview of the 12 review articles selected for inclusion; 5 were systematic reviews 16, 17, 21, 22, 29, 4 were meta‐analyses 14, 15, 18, 19 and 3 were both a systematic review and a meta‐analysis 12, 20, 33. In the 12 reviews, a total of 240 individual studies were reviewed; 64 studies were included in two or more of the 12 reviews, while 176 studies were included in only one of the 12 reviews (Table S1, Supporting Information).

Table 1

Review articles focusing on obesity or weight management and HRQoL

No.	Author, date	Type/goal of review	Studies	Key findings	Strengths	Limitations
Category 1: Relationship between weight/BMI and HRQoL (baseline/pre‐intervention)
1	van Nunen et al., 2007 14	Meta‐analysis to examine differences in baseline HRQoL among seekers of surgical treatment for obesity, seekers of non‐surgical treatment, non‐treatment‐seeking persons with obesity, general population with obesity and general population	54 cross‐sectional studies Dutch, English, French and German 1996–2006	Based on both generic and obesity‐specific measures, populations with obesity experienced reduced HRQoL ◦ For both SF‐36 and IWQOL‐Lite, the most reduced HRQoL occurred in the surgical patients ◦ Comparing patients to non‐patients, SF‐36 results varied widely by subscale, with only physical functioning showing consistently reduced HRQoL for surgical patient groups ◦ However, reduced HRQoL was found on all IWQOL‐Lite subscales for patient groups compared with non‐patients After adjustment for BMI, surgical patients still demonstrated reduced HRQoL on most SF‐36 subscales, whereas for IWQOL‐Lite differences between populations disappeared after adjustment for BMI likely due to the IWQOL‐Lite being a weight‐related measure of HRQoL	Analysis of HRQoL in 100 000 geographically diverse individuals in different populations Articles included both generic (SF‐36) and obesity‐specific (IWQOL‐Lite) HRQoL measures	The general population with obesity is heterogeneous, including those who intend to seek medical intervention, those who plan their own interventions, and those who do not intend to seek treatment There is underrepresentation of the intentionally non–treatment‐seeking population ◦ Included in only 2/54 studies
2	Ul‐Haq et al., 2013 15	Meta‐analysis to determine the relationships between BMI and physical and mental HRQoL	8 cross‐sectional studies English only 2000–2011	Physical HRQoL ◦ Individuals with higher BMI had significantly reduced physical HRQoL ◦ Clear evidence of a dose relationship across all BMI categories Mental HRQoL ◦ Only reduced among individuals classified as Class III obesity (BMI ≥ 40 kg m−2)	Population studies from Australia, Canada, England, Germany, Sweden, USA Studies included 43 086 participants Sophisticated methodology ◦ Analysed pooled estimates of weighted mean difference in PCS and MCS by BMI in reference to normal weight ◦ Determined degree of heterogeneity and assessed publication bias ◦ Applied a statistical method to reduce risk of type I errors	Limited to articles that assessed HRQoL with a single, generic measure only (SF‐36) 1 of 8 studies was based on male Veteran's Administration patients only
Category 2: HRQoL after weight loss (varied interventions and/or varied study design)
3	Carson et al., 2014 16	Systematic review to examine the effects of dietary intervention on HRQoL	24 studies of ≥12 weeks' duration, including RCTs (21) and non‐RCT prospective studies (3) English, US only 1990–2012	In most studies (88%), study participants reported improved HRQoL after dietary intervention ◦ 11/24 studies indicated changes in HRQoL were likely a result of weight loss ◦ 4/24 studies demonstrated that changes in HRQoL were independent of weight loss ◦ In 9/24 studies, it was unclear if changes in HRQoL were a result of weight loss	Articles included used both generic (QWB, SF‐36, SF‐12) and obesity‐specific (IWQOL, IWQOL‐Lite) HRQoL measures	4/24 studies used only non‐obesity‐specific disease measures (e.g. measure of heart failure) ◦ 3 of the studies using non‐obesity‐specific measures reported an improvement in HRQoL over time 13/24 studies involved short‐term follow‐ups (12–26 weeks)
4	Kroes et al., 2016 17	Systematic review to review evidence for the impact of weight/BMI change on HRQoL	20 studies, including RCTs (8), prospective comparative cohorts (5), prospective single‐arm cohorts (5), cross‐sectional (1) and retrospective analyses of RCTs (1) English, US only ≥1‐year follow‐up	Of the studies that explicitly reported on the association between HRQoL and weight change: ◦ For lifestyle approach (n = 1): there were significant correlations between weight loss and all IWQOL‐Lite subscales except work ◦ For pharmaceutical intervention (n = 1): there were larger effect sizes for greater weight reductions for all IWQOL‐Lite subscales, PCS, physical functioning, and general health ◦ For bariatric surgery at ≥1‐year follow‐up: greater weight loss showed significant correlations with vitality (n = 1), physical functioning (n = 1), PCS (n = 1), and IWQOL‐Lite Total (n = 2) SF‐36 (generic): improvements in physical aspects reported more frequently than mental/psychosocial aspects IWQOL‐Lite (obesity‐specific): improvements in all or most subscales	Unique approach: to investigate the impact of weight change on HRQoL, rather than to compare improvements in HRQoL between interventions Studies had ≥1‐year follow‐up	Included only US, English articles Studies included were heterogeneous in terms of intervention (bariatric surgery [n = 12]; lifestyle approaches [n = 7]; pharmaceutical intervention [n = 1]), as well as study design and outcomes described, making inferences about associations between weight loss and changes in HRQoL challenging Although this review included studies using any HRQoL measurement, reporting was limited to SF‐36 and IWQOL‐Lite
Category 3: HRQoL after weight loss (RCTs only)
5	Maciejewski et al., 2005 18	Meta‐analysis to estimate the effect of various weight‐loss interventions (e.g. medication, diet, exercise, commercial programme, cognitive behaviour therapy, bariatric surgery) on HRQoL in RCT studies, and a meta‐analysis of the effect of weight‐loss treatment on depressive symptoms	34 RCTs	HRQoL outcomes were not consistently improved in RCTs of weight loss ◦ Generic measures: 9/34 studies showed HRQoL improvements in ≥1 domains ◦ Obesity‐specific measures: 6/11 studies showed positive treatment effects	Articles included used both generic (GHRI, GWB, SF‐36, SIP, VAS) and obesity‐specific (IWQOL, OP) HRQoL measures Study duration varied from 6 to 208 weeks; 8 studies of >52 weeks’ duration and 26 studies of ≤1 year's duration Study quality was assessed according to 6 different criteria: concealment of randomization, blinding, loss to follow‐up, intention‐to‐treat analysis, adjustment for mediating effects of weight loss, and adjustment for multiple comparisons	Review stated that quality of studies examined was poor
6	Warkentin et al., 2013 12	Systematic review and meta‐analysis to examine the effect of weight loss (any weight‐loss intervention vs. no intervention, placebo or active comparator) on HRQoL in RCTs	53 RCTs met eligibility 11/53 included in meta‐analysis	Generic measures: 14/36 studies found significant improvements Obesity‐specific measures: 4/15 studies found significant improvements Contingency table approach (included all trials): no significant association between weight‐loss and overall HRQoL Quantitative data pooling approach (included 25% of trials): statistically significant improvements in physical but not mental health	Articles included used both generic (EQ‐5D, GHQ, QWB, SF‐36, VAS) and obesity‐specific (IWQOL‐Lite, M‐A QoLQII, OAS, OP, ORWELL, WRSM) HRQoL measures Different analytical methods were used ◦ Contingency table approach: incorporates information from all studies examined, but gives them equal weight, such that the magnitude of changes cannot be compared ◦ Conventional random effects meta‐analytical technique: considered to be more rigorous because uses study‐specific values and inverse‐variance weighting to generate pooled estimates; however, quantitative data pooling was limited to only 25% of available studies due to the poor quality of reporting	Data from most studies could not be quantitatively pooled for meta‐analysis In 35/53 RCTs, study duration was <1 year
Category 4: HRQoL after bariatric surgery
7	Magallares and Schomerus, 2015 19	Meta‐analysis to compare HRQoL before and 1 year after bariatric surgery	21 studies English, German, Italian, Portuguese and Spanish	Compared to pre‐surgery, patients scored higher in physical and mental components of the SF‐36 1 year after bariatric surgery ◦ The size of the effect was much greater for the physical component than the mental component, but both effects were very high There was a large variability/heterogeneity in amount of improvement in both PCS and MCS	Studies drawn from a diverse literature base that encompassed five languages	Study design of included studies was not indicated Limited to those with a follow‐up period of ≤1 year and those that administered only SF‐36
8	Lindekilde et al., 2015 20	Systematic review and meta‐analysis to assess the impact of bariatric surgery (15 different methods) on HRQoL and the between‐study variation	72 studies, including cohort studies (60), non‐randomized studies (5), randomized studies (7)	Bariatric surgery had a significant positive influence on HRQoL ◦ Influence was greater on physical vs. mental HRQoL ◦ Greater effects were found for obesity‐specific measures of HRQoL than for other types A large variability (heterogeneity) in HRQoL outcomes was found	Articles included used 22 HRQoL measures, including generic (EQ‐5D, SF‐36), obesity‐specific (IWQOL, IWQOL‐Lite, M‐A QoLQII, WRSM), combined generic/obesity‐specific (HRQoL‐HSP) and gastrointestinal‐specific (GIQLI) HRQoL measures Follow‐up ranged from 3 to 120 months Controlled for multiple other factors (baseline BMI, age, type of measure, type of surgery, months to follow‐up, year of publication and country of study) Provided effect sizes of changes in HRQoL	HRQoL scales and subscales were categorized into 5 domains (physical, mental, social, functional and total), but authors do not indicate how they assigned scores to the domains Majority of included studies used a non‐randomized design; none of the randomized studies used a non‐surgical control group
9	Andersen et al., 2015 21	Systematic review to study the long‐term (i.e. ≥5 years) effects of bariatric surgery (6 different methods) on HRQoL	7 prospective studies, 2 with control groups English ≥5 year follow‐up	6/7 studies showed improvements in 9 aspects of HRQoL Peak improvements in HRQoL observed during first 1–2 postoperative years (characterized by the most meaningful amount of weight loss), followed by a gradual decline that stabilized at 5 years 5‐year postoperative scores were an improvement from preoperative scores, but lower than the population norm ◦ Of the statistically significant improvements in HRQoL, 92% were clinically meaningful (i.e. >0.5 standard deviation change from baseline)	Review focuses exclusively on high‐quality, long‐term (5–10 years), prospective studies of bariatric surgery All studies included were high quality, defined as attrition rate <50%, and 90% power to detect >0.5 standard deviation change from baseline using a two‐sided paired test Articles included used both generic (15D, GHRI scale, GWB, NHPII, SF‐36) and obesity‐specific (IWQOL‐Lite, OP, Weight Distress) HRQoL measures	Attrition rates in included studies ranged from 8 to 39.2% Included studies were heterogeneous with respect to baseline BMI, HRQoL instruments and surgical methods
10	Jumbe, 2015 22	Systematic review to assess psychosocial HRQoL of bariatric surgery patients at a minimum of 1‐year follow‐up compared to: Individuals receiving non‐surgical interventions Non‐treated comparison groups	11 studies, including RCTs (2), prospective cohorts (8) and retrospective analysis (1) (SF‐36 measured at follow‐up only)	Bariatric surgery vs. non‐surgical treatment ◦ For all 3 studies using the SF‐36 scale, bariatric surgery groups had better HRQoL outcomes vs. non‐surgical interventions ◦ 1/3 studies using the SF‐36 scale reported significantly better outcomes across all subscales Bariatric surgery vs. control groups ◦ For 5 of the 7 studies assessing generic and/or obesity‐specific HRQoL pre‐and posttreatment, patients in surgical groups reported higher HRQoL vs. control groups at follow‐up ◦ Improvements were seen in all or some of the mental/psychosocial aspects of HRQoL	Articles included both generic (SF‐36, SF‐12, EQ‐5D, SIP), and obesity‐specific measures (IWQOL‐Lite, OP) Follow‐up ranged from 1 to 10 years Study quality was evaluated	2 studies assessed HRQoL in individuals with both obesity and diabetes 23, 24 4 studies 25, 26, 27, 28 used data from the Utah Obesity Study; thus, reporting these studies separately may skew the conclusions Conclusion of “persistent psychological issues post‐surgery” is an overstatement of the data presented in the reviewed studies Despite better outcomes reported in 3 of the 4 studies comparing bariatric surgery with non‐surgical treatment, authors state that, overall, these studies show moderate evidence of similar improvements in HRQoL in these 2 groups Despite reporting of improvements in mental/psychosocial HRQoL in several of the studies, authors state that long‐term psychosocial HRQoL does not improve after bariatric surgery compared to controls
11	Hachem et al., 2016 29	Systematic review to examine HRQoL as an outcome of bariatric surgery by comparing: ◦ Bariatric surgery to alternative weight‐loss interventions ◦ Different types of bariatric surgery	13 studies ◦ 7 studies (1 RCT, 6 quasi‐experimental studies): bariatric surgery vs. an alternative weight‐loss intervention ◦ 6 studies (5 RCTs, 1 quasi‐experimental study): 1 type of bariatric surgery vs. another English	Significant HRQoL improvements following bariatric surgery ◦ Greater improvements in surgical vs. non‐surgical interventions Significant HRQoL improvements in gastric bypass and laparoscopic sleeve gastrectomy vs. vertical banding gastroplasty and laparoscopic adjustable gastric banding, respectively No differences in HRQoL between variations of the same type of surgery (e.g. gastric bypass vs. mini gastric bypass)	Articles included used generic (SF‐36, M–A QoLQII, SIP, GHRI), obesity‐specific (IWQOL‐Lite, OWLQOL, WRSM, GIQLI, QOLOD) and gastrointestinal‐specific HRQoL measures Follow‐up ranged from 2 months to 10 years	2 studies measured HRQoL post‐surgery only 6 studies reported follow‐ups of ≤1 year Results were not included for the 10‐year follow‐up of Karlsson et al. 30 or the 6‐year follow‐up of Adams et al. 27 Only 2 studies evaluated between‐group differences Each comparison of different surgery types was made in only a single study Reporting of HRQoL results was inconsistent, with some reporting overall scores, some reporting composite scores, and some reporting selected subscale scores 2 included studies 31, 32 (studies by Canetti et al. in 2009 and 2013) used the same study sample
12	Driscoll et al., 2016 33	Systematic review and meta‐analysis of studies reporting HRQoL data ≥5 years after bariatric surgery and in non‐surgical control groups with obesity	9 studies, including cross‐sectional studies (7), prospective cohort study (1), and non‐randomized controlled trial (1)	Systematic review of all studies and measures ◦ Greater improvements were noted in both the physical and mental domains of HRQoL for the surgical groups compared with the control groups; however, there were inconsistencies in results (i.e. favouring surgical group, favouring control group, and no difference) in both physical and mental domains Meta‐analysis of SF‐36 results ◦ Inconsistencies that had been seen in the systematic review were not seen in the meta‐analysis ◦ Significant improvement in all mental domains after 5 years favouring the surgical group compared with the controls; and 3/4 physical domains The magnitude of improvement in surgical groups vs. control groups was greater for the physical than mental domains	Articles included used both generic (SF‐36, EQ‐5D, Current Health Scale from the GHRI and obesity‐specific (IWQOL‐Lite, OP) measures Studies had follow‐up periods of 5 to 25 years	Meta‐analysis could only be conducted on studies reporting SF‐36 scores

HRQoL‐HSP is sometimes referred to as ‘Lewin‐TAG’.

15D, 15‐dimensional measure; EQ‐5D, EuroQuol‐5D measure; GHQ, General Health Questionnaire; GHRI, General Health Rating Index; GIQLI, Gastrointestinal Quality of Life measure; GWB, General Well‐Being measure; HRQoL, health‐related quality of life; HRQoL‐HSP, Health‐related quality of life‐Health State Preference Assessment; IWQOL, Impact of Weight on Quality of Life; IWQOL‐Lite, Impact of Weight on Quality of Life‐Lite; M–A QoLQII, Moorehead–Ardelt Quality of Life Questionnaire II; NHP, Nottingham Health Profile; OAS, Obesity Adjustment Survey; OP, Obesity‐related Problems scale; ORWELL, Obesity‐Related Well‐Being questionnaire; OWLQOL, Obesity and Weight Loss Quality of Life; QOLOD, Quality of Life, Obesity and Dietetics Rating Scale; QWB, Quality of Well‐Being scale; RCT, randomized controlled trial; SF‐36, Medical Outcomes Study Short‐Form‐36; SIP, Sickness Impact Profile; VAS, Visual Analogue Scale; WRSM, Weight‐Related Symptoms Measure.

Some of the review articles provided a definition of QoL and/or HRQoL 12, 16, 20, 21, 29. Definitions were similar and confirmed HRQoL as a multidimensional concept of the individual's perceptions of the impact of their disease on an individual level. All reviews aimed to analyse HRQoL in relation to weight, and could be logically organized into four categories (Table 1). Following a brief discussion of the assessment methods used in all included reviews, the reviews are described by category, below.

Assessment of health‐related quality of life

Among the 12 review articles examined, a total of 23 different measures of HRQoL were identified (Table S2) and could be categorized as follows: 11 generic, 10 obesity‐specific, 1 combined generic/obesity‐specific and 1 gastrointestinal‐specific. We did not report results of measures of depression, anxiety, stress or emotions 12, 18, 22, nor did we include any non‐obesity‐specific disease measures (e.g. Minnesota Living with Heart Failure Questionnaire, Pelvic Floor Impact Questionnaire), as these measures are not true assessments of HRQoL 34. We have also not reported results obtained from the Bariatric Analysis and Reporting Outcome System (BAROS) 35, as it can only be applied to post‐surgical (not pre‐ vs. post‐surgical) outcomes for bariatric surgery patients.

A total of 10 of the 12 review articles included studies that used both generic and obesity‐specific measures of HRQoL 12, 14, 16, 17, 18, 20, 21, 22, 29, 33, whereas 2 of the review articles included only generic measures 15, 19. The most commonly used measures were the generic Medical Outcomes Study Short‐Form‐36 (SF‐36) 36 and the obesity‐specific Impact of Weight on Quality of Life‐Lite (IWQOL‐Lite) questionnaires 37. A total of 5 of 12 the reviews specified that they included both generic and obesity‐specific measures in order to provide a more complete understanding of the patient populations 12, 17, 18, 20, 29.

Category 1: Relationship between weight/body mass index and health‐related quality of life (baseline/pre‐intervention)

A total of 2 of 12 reviews examined the association between weight/body mass index (BMI) and HRQoL without, or prior to, any intervention 14, 15. van Nunen et al. performed a meta‐analysis of cross‐sectional differences among five subgroups, including: non‐patient subgroups (general population, general population with obesity and non‐treatment‐seeking people with obesity) and patient subgroups (non‐surgical weight‐loss patients and bariatric surgery patients) 14 (Table 1). For both SF‐36 and IWQOL‐Lite, the greatest reduction in HRQoL was observed in the surgical patients. SF‐36 results varied widely by subscale, with only a consistent reduction in the physical functioning HRQoL subscale for patient vs. non‐patient groups.

However, reduced HRQoL was found on all IWQOL‐Lite subscales for patient (compared with non‐patient) groups. After adjustment for BMI, reduced HRQoL on most SF‐36 subscales remained for surgical patients, whereas for IWQOL‐Lite differences between general, non‐treatment, conservative treatment and surgical treatment subgroups disappeared, suggesting that when using the IWQOL‐Lite questionnaire in cross‐sectional analyses, body weight is the main determinant of HRQoL.

Ul‐Haq et al. evaluated HRQoL across studies that used SF‐36 and cross‐sectional data from six countries (Table 1) 15. In this well‐designed meta‐analysis, the authors analysed pooled estimates (totalling 43 086 participants) of the weighted mean difference in the Physical Component Summary (PCS) score and the Mental Component Summary (MCS) score with reference to normal weight, determined the degree of heterogeneity, assessed publication bias and applied a statistical method to reduce the risk of Type I errors. Compared with normal‐weight adults, those with higher BMI (≥25 kg m−2) had significantly reduced PCS scores and a dose relationship was evident across all BMI categories. In contrast, compared with the normal‐weight adults, MCS scores were significantly reduced only in those with Class III obesity (BMI ≥ 40 kg m−2), but not Class I (BMI 30–34.9 kg m−2) or Class II (BMI 35–39.9 kg m−2) obesity 15.

Category 2: Health‐related quality of life after weight loss (varied interventions and/or study design)

Carson et al. 16 assessed changes in HRQoL after dietary interventions in 21 randomized controlled trials (RCTs) and 3 non‐RCTs (Table 1) 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61. Twenty‐two studies reported improvements in HRQoL over time; 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 52, 53, 54, 56, 57, 58, 59, 60, 61 however, 3 of these 22 studies used only non‐obesity‐specific disease measures to assess HRQoL 38, 44, 54. Mean weight loss ranged from 2 to 10 kg for the groups receiving active treatment (i.e. calorie restriction, healthy diet or commercial programme).

Four studies demonstrated that changes in HRQoL were independent of weight loss; 39, 40, 41, 42 11 studies indicated that changes in HRQoL were probably a result of weight loss 38, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52 (three of these included only a non‐obesity‐specific measure of disease impairment); 38, 44, 51 and in the remaining nine studies, the role of weight loss in HRQoL changes was unclear 53, 54, 55, 56, 57, 58, 59, 60, 61.

Kroes et al. assessed the impact of weight/BMI change on HRQoL (SF‐36, IWQOL‐Lite, or both) in 8 RCTs and 12 non‐RCTs following a number of different interventions, including lifestyle approaches (exercise, dietary weight loss or counselling), pharmaceutical intervention and bariatric surgery 17. For the seven studies that reported the association of weight loss and HRQoL, one study of lifestyle change resulted in a median weight loss of 0.36 kg; for the one pharmaceutical intervention, mean weight loss was 2.7%; for the studies of bariatric surgery with ≥1‐year follow‐up, percent weight reduction was 20–38.8% and percent excess weight loss was 56.4–62.7%. Regardless of intervention, improvements in the physical aspects of the SF‐36 were reported more frequently than improvements in mental/psychosocial aspects, whereas improvements in all or most of the obesity‐specific IWQOL‐Lite subscales were demonstrated. However, a significant degree of heterogeneity in the included studies precluded drawing conclusions about the specific association between weight/BMI change and HRQoL.

Category 3: Health‐related quality of life after weight‐loss (randomized controlled trials only)

Two reviews described HRQoL outcomes after RCTs of weight‐loss interventions, one published in 2005 18 and the other in 2013 12 (Table 1). The earlier review included 34 studies, while the later one included 53 studies (with 14 overlapping studies; see Table S1). As well as an increase in the number of studies, the quality of reporting of HRQoL results improved over time, allowing more sophisticated analyses in the more recent review.

One similarity between these two reviews is that weight‐loss interventions were quite diverse. Although both reviews included meta‐analyses, the 2005 review 18 − due to poor quality of reporting and insufficient data on measures of HRQoL − used a meta‐analysis only on the effect of weight‐loss treatment on depressive symptoms, whereas in the 2013 review 12 meta‐analyses were used to examine the relationship between changes in weight and HRQoL.

In the studies reported in the 2005 review 18, 9 of 34 RCTs demonstrated improvements in generic HRQoL in one or more domains, although the domains varied by study. Six of the 11 that used obesity‐specific measures showed positive treatment effects. Weight loss was not reported in this review. In the studies reported in 2013 by Warkentin et al. 12, 14 of 36 studies reporting generic measures of HRQoL found significant improvements, whereas 4 of 15 studies reporting obesity‐specific measures found significant improvements. Weight loss varied from ≤5 to ≥10% (specific weight loss values were not reported). Conclusions from both of these reviews of RCTs were similar. Maciejewski et al. 18 concluded: ‘HRQoL outcomes, including depression, were not consistently improved in RCTs of weight loss.’ Warkentin et al. 12 concluded: ‘Certainly, compelling and definitive RCT‐level data to support the notion that HRQoL is consistently and robustly improved following weight loss is not available.’

Category 4: Health‐related quality of life after bariatric surgery

Six reviews examined HRQoL after bariatric surgery (Table 1) 19, 20, 21, 22, 29, 33. Inclusion/exclusion criteria differed between the reviews, resulting in eight overlapping studies between Lindekilde et al. 20 and Magallares and Schomerus 19, six between Lindekilde et al. 20 and Andersen et al. 21, four between Lindekilde et al. 20 and Hachem et al. 29, and three between Lindekilde et al. 20 and Jumbe et al. (Table S1) 22. The remaining review combinations had two or fewer overlapping studies. Each of these reviews of bariatric surgery outcomes has its own strengths and limitations (Table 1).

The six reviews reported and/or analysed HRQoL data using different methods, making a comparison of the results challenging. Only the reviews by Andersen et al. 21, Lindekilde et al. 20, and Magallares and Schomerus 19 reported results in terms of standardized effect sizes (i.e. the magnitude of the pre‐ to post‐surgery changes were reported in a standardized fashion). Reporting standardized effect sizes allows direct comparisons across studies. Nevertheless, the specific methods used to calculate effect sizes varied across reviews, making comparisons less meaningful.

Additional inconsistencies in HRQoL reporting were observed across studies. In Andersen et al. 21, primary outcomes (defined as summary scores of generic or obesity‐specific measures or a measure of overall health status or overall well‐being) and secondary outcomes (defined as domain scores, such as social interaction, physical appearance and self‐regard) were reported separately to minimize multiple comparisons. In the review by Lindekilde et al. 20, all of the HRQoL scales and subscales were categorized into five domains: ‘physical’, ‘mental’, ‘social’, ‘functional’ and ‘total’, together with ‘overall HRQoL’ and ‘HRQoL for all measures’. However, no information was provided on how the various subscales of each assessment measure were assigned to these domains, nor was the distinction between total HRQoL and overall HRQoL described, making their analyses difficult to replicate and results difficult to interpret. Finally, reporting of study results was inconsistent in the reviews by Hachem et al. 29 and Jumbe et al. 22, with some reporting overall scores, some reporting composite scores and some reporting subscale scores, also making results difficult to interpret.

Weight loss was reported differently across bariatric surgery reviews, as well as inconsistently within any single review. Study duration ranged from 5 to 72 months. For reviews reporting change in BMI from baseline to end of study 20, 21, 22, 33, BMI change ranged from 4.6 to 30.6 kg m−2. For reviews reporting mean weight loss in kg 22, weight reduction was 13.5−45.1 kg. Two reviews 19, 29 did not report any end‐of‐trial values for change in BMI or weight.

Medical Outcomes Study Short‐Form‐36 (SF‐36) results of bariatric surgery studies

All bariatric surgery reviews included SF‐36 results. Magallares and Schomerus 19 found a large variation in the degree of improvement in both PCS and MCS at ≤1‐year post‐surgery. Although effect sizes associated with these changes were much greater for the physical component than the mental component, both effects were very large.

Although Lindekilde et al. 20 did not report results separately for PCS and MCS, they examined differences between ‘physical and mental domains on the SF‐36’ adjusted for baseline scores, reporting that these differences were significantly greater for the physical than for the mental domains. In addition to the SF‐36 analysis, these authors computed physical and mental scores from the various measures administered. The effect size for physical HRQoL was significantly greater than that for mental HRQoL and this difference remained significant, even after adjusting for other factors (baseline BMI, age, type of measure, type of surgery, months to follow‐up, year of publication and country in which the study was carried out).

The review by Jumbe et al. 22 revealed significantly better SF‐36 outcomes in three out of three studies for people who had undergone bariatric surgery compared with those receiving non‐surgical interventions. One out of the three RCTs reported significantly better outcomes across all SF‐36 subscales. Five out of seven studies in which SF‐36 outcomes after bariatric surgery were compared with those obtained in non‐treated control groups showed greater improvements in the surgery group compared with the controls, although there was some overlap among the patient populations included in these studies (see Table 1).

Four of the six studies reporting SF‐36 results in the review by Hachem et al. 29 reported a significant improvement in the physical aspects of HRQoL and three of the studies reported a significant improvement in the mental aspects of HRQoL after bariatric surgery, whereas there were no significant changes in these domains in the non‐surgical comparator groups in two of the studies.

Two of the reviews included herein comprised only long‐term studies (≥5‐year duration) 21, 33. In the review by Andersen et al. 21, two of the three studies using the SF‐36 showed significant improvements in both PCS and MCS at 5–6 years; 25, 62 the remaining study showed significant improvements only in PCS 63. Peak improvements in PCS occurred at 1–2 years across all three studies. MCS scores showed a less consistent pattern. While the systematic review conducted by Driscoll et al. 33 showed some inconsistencies in SF‐36 results, meta‐analysis results revealed significant improvements in both physical and mental health domains of the SF‐36 after 5 years, favouring surgical over control groups. Improvements were greater in the physical compared with the mental domains.

In general, results of SF‐36 scores following bariatric surgery showed improvements in HRQoL relative to non‐surgical groups. Moreover, although these changes were seen in both mental and physical domains, improvements in physical domains of HRQoL seem to be greater.

Obesity‐specific health‐related quality of life results of bariatric surgery studies

Five of the six reviews included obesity‐specific HRQoL results 20, 21, 22, 29, 33. In the three studies 25, 30, 64 reviewed by Andersen et al. 21 that used an obesity‐specific measure of HRQoL, significant and very large effects were reported at 5–10 years. In addition, effect sizes were generally larger for obesity‐specific than for generic measures. Similarly, effect sizes were greater for obesity‐specific than for generic measures of HRQoL in the Lindekilde et al. 20 review.

Two studies included in the review by Hachem et al. reported between‐group differences on obesity‐specific measures of HRQoL 26, 65. Specifically, in the study by Adams et al. 26, significant improvements were found on the IWQOL‐Lite questionnaire in the post‐gastric bypass surgery vs. two comparator groups (people with obesity seeking bariatric surgery who did not have surgery and people with obesity who were not seeking surgery).

In the Brunault et al. 65 study, both laparoscopic adjustable gastric banding (LAGB) and laparoscopic sleeve gastrectomy (LSG) resulted in improved physical, psychosocial, sexual and diet experience domains on the QoL, Obesity and Dietetics Rating Scale vs. pre‐operative assessments. In addition, the ‘comfort with food scale’ showed greater improvement in the LSG group compared with the LAGB group at 6, but not at 12, months.

In the review by Driscoll et al. 33, three studies 30, 66, 67 included the Obesity‐related Problems scale (OP scale) and one study 25 included the IWQOL‐Lite. An observational, cross‐sectional study by Raoof et al. 66 indicated that OP scores for patients who had undergone gastric bypass surgery an average of 11.5 ± 2.7 years earlier showed improved HRQoL vs. scores obtained by a matched group awaiting bariatric surgery; however, OP scores for the post‐surgical group also revealed that HRQoL remained impaired when compared with the general population. In the study by Karlsson et al. 30, the OP scale improved significantly for bariatric surgery patients at 10 years vs. pre‐surgery and these improvements were greater than those seen in patients treated with non‐surgical interventions. In the study by Aftab et al 67,. OP scores obtained in post‐bariatric surgery patients at 5 years were superior to those obtained in patients pre‐surgery. Kolotkin et al. 25 found that patients who had undergone gastric bypass surgery 6 years previously had significantly greater improvements from baseline in all five IWQOL‐Lite domains compared with people with obesity who sought but did not receive gastric bypass, and also compared with people with severe obesity in the general population.

In summary, using obesity‐specific HRQoL measures, positive effects on HRQoL were seen following bariatric surgery, and these changes tended to be greater than those seen with SF‐36.

Discussion

This is the first systematic review of reviews to synthesize published information on the impact of obesity and weight loss on HRQoL. By including only meta‐analyses and systematic reviews, and excluding narrative reviews, bias was minimized. Moreover, by reviewing multiple reviews, we were able to integrate information from a large number of studies: only 64 (27%) of the 240 studies were included in more than one review, and this small degree of overlap limited the risk of duplication of conclusions, while ensuring the reviews were not overly selective. This systematic review of reviews illustrates the significant and negative impact that overweight/obesity has on HRQoL, regardless of study population 14, 15.

In reviews examining the relationship between HRQoL and obesity in various populations without or prior to intervention, increased obesity was associated with decreased HRQoL, particularly in those with Class III obesity (BMI ≥ 40 kg m−2) 15 and those seeking bariatric surgery 14, suggesting that low HRQoL may provide motivation in these patients to undergo an invasive intervention. Furthermore, the physical aspects of HRQoL seem to be more closely associated with degree of obesity than the mental aspects of HRQoL; in population data from six countries the relationship between BMI and SF‐36 PCS was dose‐dependent, with poorer physical HRQoL occurring in those with higher BMIs, whereas MCS scores were reduced only in those with Class III obesity.

Reviews limited to RCTs 12, 18 represent the most robust dataset available, and although one might expect consistent associations between weight loss and improved HRQoL, these studies are not conclusive, despite inclusion of both generic and obesity‐specific data. In the two reviews limited to RCTs, one reported significant improvements in one or more domains of generic HRQoL in 9 out of 34 included RCTs 18, whereas the other review reported improvements in 14 out of 36 included studies evaluating HRQoL using a generic measure 12. Obesity‐specific HRQoL results were also inconsistent, with one review reporting improved obesity‐specific HRQoL in 6 out of 11 included studies 18 and the other reporting improved obesity‐specific HRQoL in 4 out of 15 included studies 12.

The lack of consistently demonstrated associations between weight loss and improved HRQoL in RCTs may be due to the following factors: diverse HRQoL measures have been used; weight‐loss interventions have been heterogeneous, with some more successful at inducing weight loss than others; studies may have been underpowered to detect differences in HRQoL outcomes; and there has been poor reporting of HRQoL outcomes (since weight loss, rather than HRQoL, is usually the primary outcome). To bring more clarity to these results, authors of two of the systematic reviews recommend that future research focus on prospective, long‐term studies, especially RCTs or large, well‐designed, observational studies with high retention rates, comparator groups and carefully chosen generic and obesity‐specific HRQoL measures 12, 21.

Results from reviews that include studies of mixed design (RCTs and non‐RCTs) and/or mixed interventions are even more difficult to interpret 16, 17, especially when treatments are diverse (e.g. bariatric surgery, pharmaceutical interventions and lifestyle treatments) and result in vastly divergent weight loss outcomes 17. Overall, more randomized, controlled, high‐quality studies are needed to better understand the relationship between the various possible weight‐loss interventions and changes in HRQoL.

When reviews were limited to those undergoing bariatric surgery, there was less variation in results, with post‐bariatric surgery patients demonstrating improvements in both physical and mental aspects of generic HRQoL. It is likely that the greater weight reductions seen with bariatric surgery, compared with dietary, medical and lifestyle treatments, contribute to the consistency of improvements in HRQoL. In addition, baseline HRQoL scores are more impaired for people undergoing bariatric surgery than for patients receiving non‐surgical treatments 14 allowing the possibility of greater improvement. Improvements in physical aspects were greater and reported more consistently than improvements in mental aspects of HRQoL when assessed with the SF‐36. Comparison of post‐surgical effect sizes indicated consistently larger effects with obesity‐specific measures compared with generic measures, suggesting that obesity‐specific measures may be more sensitive to change.

That such effects were generally more marked when obesity‐specific measures were utilized is consistent with recommendations made in several reviews that both types of measures be incorporated into future studies 12, 17, 20, 21, 29.

In the reviews comparing HRQoL outcomes in patients who have undergone weight‐loss surgery with non‐surgical comparator groups, improvements in HRQoL were generally greater in the surgical groups, despite some inconsistency of findings. Taken as a whole, these findings suggest the value of assessing HRQoL in patients with clinically severe obesity so that they may receive the HRQoL benefits from bariatric surgery.

It is encouraging that weight loss is often associated with improvements in HRQoL in people with obesity. Nevertheless, some reviews demonstrated significant variability in HRQoL after weight‐loss intervention. Articles included in this systematic review of reviews cited several potential limitations, including insufficient data due to patient dropout and lack of follow‐up, and a shortage of studies reporting on obesity‐specific HRQoL. For example, in the review by Warkentin et al. 12, only 25% of studies using SF‐36 could be included in the quantitative data pooling due to poor reporting quality. This low quality of reporting and lack of data may be because HRQoL is usually a secondary outcome in studies. Another limitation in the included reviews of weight‐loss studies is that length of follow‐up varied considerably, some with durations as short as 6 weeks, making it difficult to draw meaningful conclusions about longer‐term outcomes. The majority of studies had a short‐term (<24 months) or medium‐term (24–60 months) follow‐up period and only two reviews (both evaluating the effects of bariatric surgery on HRQoL) focused exclusively on studies with long‐term follow‐up periods (≥5 years) 21, 33.

The assessment of the quality of studies included in a review is also an important indicator of the strength of the conclusions of the authors; yet, only approximately 40% of systematic reviews evaluate the quality of included studies 68.

Notably, in this systematic review of reviews, a higher number (9 of the 12 review articles [75%]) concerned themselves with study quality, either by including a direct assessment of study quality 12, 16, 17, 18, 22, 33 or by using ‘acceptable quality’ (variably defined) as selection criteria for study inclusion 20, 21.

We also noted differences across the 12 reviews with respect to HRQoL measures, country and language of the studies, reporting of results, method for determining effect sizes, length of follow‐up, type of weight‐loss intervention and population being studied. We made every effort to describe similarities and differences among reviews and to report the number of reviews that made particular conclusions. We believe our review represents a true synthesis of current and diverse reviews in this field and, as such, has broad applicability.

We make no attempt to draw conclusions about the HRQoL of specific patient populations/subgroups (e.g. different ethnicities/cultures, age or comorbidities). Studies not discussed in this review have shown that these variables may be helpful in predicting the impact that weight and/or weight loss have on HRQoL 69, 70, 71, 72. For example, increasing age in people with overweight or obesity is associated with increased impairment in some domains of HRQoL, such as physical function, sexual life and work, but not others 72. Some studies conclude that women with overweight or obesity report more impairment than men in some HRQoL domains 71, 72. Furthermore, HRQoL in people with obesity varies depending on presence of comorbidities 69, 70. Two reviews 15, 17 recommend that future studies assess the impact of comorbidities on HRQoL.

Four reviews 12, 16, 17, 18 recommended exploring mediators of changes in HRQoL to understand if the driver of HRQoL improvements is weight loss itself, the weight‐loss intervention or changes in other variables. A recent study 73 used multiple mediation analysis to explore causal mechanisms underlying the relationship between weight loss and improved HRQoL in two weight‐loss trials investigating phentermine/topiramate. Results indicated that improved HRQoL was primarily mediated by weight loss, but decreased depressive symptoms also accounted for improvements in HRQoL.

Similarly, a mediation analysis of three RCTs with liraglutide 3.0 mg showed that improvements in IWQOL‐Lite total and physical function scores in patients with type 2 diabetes were primarily driven by weight loss 74. Several other recommendations for future research were provided in the included reviews (see Fig. 3 for summary).

Figure 3

Summary of recommendations for future studies. BMI, body mass index; HRQoL, health‐related quality of life; PCS, Physical Component Summary. [Colour figure can be viewed at wileyonlinelibrary.com]

After conducting this systematic review of reviews, we believe the next wave of research studies should focus on the interactions between HRQoL and other variables, such as gender, fitness level, comorbidities or body image. For example, do the associations between HRQoL and obesity and/or weight loss consistently vary by other variables and how do these variables interact with each other? Is there a gradient in HRQoL by weight‐loss treatment, such that treatments inducing greater weight loss are or are not associated with greater HRQoL improvements? How much weight loss is needed for improvement in HRQoL and what is the impact of weight regain on HRQoL? Is improvement in HRQoL after weight‐loss treatment dependent primarily on amount of weight loss, or do other factors have a role? Can we predict which individuals will experience improved HRQoL after weight loss?

Inclusion of HRQoL assessments, especially obesity‐specific assessments, in regular patient evaluations could guide the development of broad healthcare policies that recognize the bio‐psychosocial impact of the growing obesity epidemic. HRQoL assessments can help patients and providers differentiate between treatments that have similar weight‐loss patterns but different side‐effect profiles or different impacts on HRQoL. It is desirable to reach a consensus for the best way to evaluate the impact of obesity and/or weight loss on HRQoL in future studies to facilitate comparison of results across studies 77.

In summary, this is the first systematic review of reviews to synthesize research on the impact of obesity and weight loss on HRQoL. We found that obesity was associated with significantly lower HRQoL in all populations. We also established an important relationship between weight loss and improved HRQoL, which is demonstrated most consistently following bariatric surgery, but less consistently following non‐surgical weight‐loss interventions.

In order to build upon these findings, we recommend longer‐term studies that use both generic and obesity‐specific measures and that evaluate the impact of other factors (e.g. comorbidities, fitness level, body image) on HRQoL after various weight‐loss interventions. These findings, plus those of future studies and review articles, will help us better understand the complex and multifaceted nature of obesity and its impact on the daily lives of our patients.

Conflict of Interest Statement

RLK receives royalties from Duke University as IWQOL‐Lite developer; support as a consultant to Eisai, Janssen and Novo Nordisk; and grant support from University of Utah (Grant DK055006; the National Institutes of Health/The National Institute of Diabetes and Digestive and Kidney Diseases). JRA has no conflicts to declare.

Author contributions

Both RLK and JRA were responsible for the content and organization of the manuscript, as well as the search parameters of this systematic review of reviews. RLK did the majority of the writing of the manuscript, with JRA contributing significantly to the discussion and recommendations. Both authors take full responsibility for the content of this manuscript.

Table S1. Overlap of studies included in the reviews.

Table S2. Health‐related quality‐of‐life measures.

Click here for additional data file.