Skip to main content
Download PDF

Understanding the risk of developing weight-related complications associated with different body mass index categories: a systematic review

Diabetology & Metabolic Syndrome volume 14, Article number: 186 (2022) Cite this article

Abstract

Background

Obesity and overweight are major risk factors for several chronic diseases. There is limited systematic evaluation of risk equations that predict the likelihood of developing an obesity or overweight associated complication. Predicting future risk is essential for health economic modelling. Availability of future treatments rests upon a model’s ability to inform clinical and decision-making bodies. 

This systematic literature review aimed to identify studies reporting (1) equations that calculate the risk for individuals with obesity, or overweight with a weight-related complication (OWRC), of developing additional complications, namely T2D, cardiovascular (CV) disease (CVD), acute coronary syndrome, stroke, musculoskeletal disorders, knee replacement/arthroplasty, or obstructive sleep apnea; (2) absolute or proportional risk for individuals with severe obesity, obesity or OWRC developing T2D, a CV event or mortality from knee surgery, stroke, or an acute CV event.

Methods

Databases (MEDLINE and Embase) were searched for English language reports of population-based cohort analyses or large-scale studies in Australia, Canada, Europe, the UK, and the USA between January 1, 2011, and March 29, 2021. Included reports were quality assessed using an adapted version of the Newcastle Ottawa Scale.

Results

Of the 60 included studies, the majority used European cohorts. Twenty-nine reported a risk prediction equation for developing an additional complication. The most common risk prediction equations were logistic regression models that did not differentiate between body mass index (BMI) groups (particularly above 40 kg/m2) and lacked external validation. The remaining included studies (31 studies) reported the absolute or proportional risk of mortality (29 studies), or the risk of developing T2D in a population with obesity and with prediabetes or normal glucose tolerance (NGT) (three studies), or a CV event in populations with severe obesity with NGT or T2D (three studies). Most reported proportional risk, predominantly a hazard ratio.

Conclusion

More work is needed to develop and validate these risk equations, specifically in non-European cohorts and that distinguish between BMI class II and III obesity. New data or adjustment of the current risk equations by calibration would allow for more accurate decision making at an individual and population level.

Background

Obesity is a disease in its own right but also a major risk factor for several chronic diseases, including type 2 diabetes (T2D), cardiovascular (CV) disease (CVD), particularly heart disease and stroke, musculoskeletal disorders, fatty liver disease, hypertension, and some cancers [1]. Obesity is therefore strongly associated with increased risk of morbidity and mortality in affected individuals. Globally, obesity prevalence is rising with a close to tripling of the worldwide prevalence between 1975 and 2016, bringing the number of adults with overweight to 1.9 billion, of whom over 650 million had obesity [2].

Obesity and overweight are defined as abnormal or excessive fat accumulation causing a deterioration in health [3,4,5]. Epidemiologically, obesity is also classified by body mass index (BMI), a simple index derived from weight (kg) and height (m2). A BMI between 25 and 29.9 kg/m2 is classified as overweight and obesity as a BMI greater than or equal to 30 kg/m2, further subdivided into class I (30.0–34.9 kg/m2) and class II (35.0–39.9 kg/m2) [2]. A BMI greater than or equal to 40 kg/m2 is referred to as class III obesity.

Furthermore, obesity imposes an increasing economic burden on society and exacerbates the costs of both obesity-related and unrelated comorbidities [6]. This is especially true of class III obesity which has a disproportionately large burden on the healthcare system [7], and in 2013 the healthcare burden of individuals with a BMI above 35 kg/m2 was 60% of total obesity-related costs [8]. By 2030, obesity above a BMI of 35 kg/m2 is predicted to rise to one in four adults in the USA, becoming the leading BMI category for certain populations, including women, non-Hispanic Black adults, and low-income adults [8, 9]. Direct healthcare and indirect economic cost burden are forecast to be upwards of tens of billions of dollars per year by 2030 [10].

Overweight, a precursor to obesity, significantly increases the risk of developing many complications also related to obesity [11]. The overweight and obesity class I–III cut-offs were developed in Western populations but vary depending on ethnicity [12], for example the original class I, BMI 30–34.9 kg/m2, has been lowered for Asian and South Asian populations to BMI 27.5 kg/m2 or 25 kg/m2 [13, 14]. To consider the health consequences of overweight while potentially distinguishing them from those that are not the result of excess adipose tissue, the patient population with a BMI of 25–29.9 kg/m2 was included in this systematic literature review (SLR) population criteria when a weight-related complication was present.

Very few analyses of populations with class III obesity exist, although a BMI of 40–59 kg/m2 is associated with substantially elevated rates of total mortality due to obesity complications [15]. The health and economic burden caused by increasing BMI have resulted in adiposity being included among the World Health Organization’s (WHO) global action plan for the prevention and control of non-communicable diseases. This action plan includes a specific objective to halt the rise in obesity prevalence at its 2010 level by 2025 [16, 17]. Controlling the growth of populations with obesity is especially relevant during global health emergencies, for example the coronavirus disease 2019 pandemic, where obesity complications are a major risk factor for survival [18, 19].

Risk equations can be developed to predict the likelihood of a person with or without a pre-existing condition of developing an associated obesity complication. These equations can help decision making at the patient and health management level. Informative risk equation outcomes hinge on choosing relevant risk predictors. Risk predictors can be lifestyle based (e.g. smoking, and exercise), socio-economic (e.g. income, and education), phenotypic (e.g. BMI, and sex), endotypic (e.g. hypertension, and insulin sensitivity), as well as health and disease related (e.g. diabetes, and heart attack). Risk predictor usage varies between equations and BMI’s association with the aforementioned diseases make it an obvious risk predictor choice to determine health outcome risk.

Greater understanding is required as to how risk equations can be used to reliably predict individual and population-based risk [20,21,22]. For example, in order to assess the risk of obesity-associated chronic diseases (such as CVD and T2D), multiple risk equations have been developed, such as the Cambridge Risk Score, QDiabetes, Finnish Diabetes Risk Score (FINDRISC), Leicester Risk Assessment, QD Score, (T2D) Framingham, and QRISK2 (CVD). Yet they lack sensitivity or validation for individuals with high BMI levels, although health outcomes are differentially impacted dependent on the specific BMI level [23, 24].

Weight related risk prediction equations that account for the statistical interaction between BMI levels and other risk predictors with the risk of outcome are of high value, without which risks associated with certain populations will not be accurately predicted in models. In the absence of sensitivity or validation for individuals with high BMI levels, a calibration (to estimate uncertain parameters and more accurately defining model uncertainty, by comparing model outputs with empirical data) of risk equations using current risk estimates from the literature (such as of the occurrence of CVD and development of T2D in high BMI individuals) may lend precision to estimates of health outcomes. We sought to identify reliable risk prediction usage that distinguishes between, and were robust for, all BMI levels.

While some studies have sought to predict the risk of weight-related mortality or complication development using risk equations, there has been limited systematic evaluation of these equations. This SLR had two overarching aims, firstly to identify published risk equations for developing obesity and overweight with a weight-related complication (OWRC) associated complications. Second, to identify studies that calculated absolute or proportional risk of individuals with OWRC, obesity, or severe obesity developing specific disease outcomes (e.g., T2D, CV event, or death). This insight will aid economic modelers in developing more accurate predictions of weight-related disease outcomes, which will in turn aid decision makers in evaluating measures that promise to improve health outcomes and reduce the social and economic burden in these populations.

Methods

A systematic review was undertaken to identify all studies that provide information on the risk associated with people living with obesity or OWRC developing complications. This review was conducted according to the York Centre for Reviews and Dissemination guidelines for conducting reviews [25] and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [26, 27]. The protocol was pre-registered on the PROSPERO register of systematic reviews (CRD42021245324).

Study identification

The following electronic databases were searched as standard evidence sources: MEDLINE, MEDLINE In-Process and e-publications ahead-of-print (via PubMed), Embase (via Embase.com), the technology appraisal submissions to the National Institute for Clinical Excellence (NICE) using risk equations, and bibliographic reference lists from included studies.

Searches were conducted on March 25, 2021. All peer-reviewed articles or errata published in the past 10 years (January 2011–March 2021) in Australia, Canada, Europe, the UK, and the USA were included. Only English language studies were included, full inclusion and exclusion criteria are reported in Additional file 1: Table S1. The search strings can be found in Additional file 1: Tables S2 (Embase) and Additional file 1: Table S3 (PubMed) (see Additional file 1).

Included studies reported on adults of ≥ 18 years with a BMI ≥ 30 kg/m2 or BMI ≥ 25 kg/m2 with the presence of at least one weight-related complication. There were no restrictions for interventions or comparators. Studies reported at least one of the following four outcomes: (1) the risk equation which calculated the risk of developing at least one additional complication: prediabetes, T2D, CVD, heart disease (including heart failure with preserved ejection fraction), acute coronary syndrome (unstable angina or myocardial infarction [MI]), stroke, musculoskeletal disorders, knee replacement/arthroplasty, obstructive sleep apnea in a population with obesity or OWRC; (2) absolute risk or proportional risk of developing T2D in a population with obesity and prediabetes or normal glucose tolerance (NGT); (3) absolute risk or proportional risk of having a CV event in a population with class III obesity and NGT or T2D; (4) absolute risk or proportional risk of mortality from knee surgery, stroke, or acute CV events in a population with obesity or OWRC. For outcomes 2–4: complications are T2D, hypertension, dyslipidemia, obstructive sleep apnea, or CVD; absolute risk refers to risk, rate, odds, and hazard outcomes; proportional risk refers to risk ratio, odds ratio, hazard ratio, relative risk, and risk reduction.

Study selection was conducted independently in duplicate. Disagreement between the reviewers was reviewed by a senior researcher (PGF), who made the final decision on inclusion or exclusion of the record. Quality appraisal and data extraction (extraction categories in Additional file 1: Table S4) were undertaken by one reviewer (ABT) and checked by a second (ER).

A thematic analysis of the evidence was completed, and results are described in a narrative summary of the evidence. Outcome data was grouped where possible to enable descriptive analysis, e.g., patient characteristics and comorbidities. Heterogeneity observed between studies, in terms of methodology and the use of definitions or descriptors, was recorded and accounted for.

Quality assessment

Studies were quality assessed based on three categories outlined by the Newcastle Ottawa Scale (NOS) [28] (with modification; categories were selection, comparability, outcome), as well as an additional category (risk equation validation) (see Additional file 1: Table S5). Between zero and fifteen stars were allocated to each study (a maximum of eight for the selection category, two for the comparability category, three for the outcome category, two for the validation category). Studies were ranked as high (≥ 13 stars), medium (11–12 stars), or low quality (≤ 10 stars).

Results

Study selection

The details of the study selection are presented in a PRISMA flow diagram in Fig. 1.

Fig. 1
figure 1

PRISMA flow diagram of study selection process. BMI body mass index, PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analysis

Full size image

A total of 3155 potentially relevant studies were identified for review. A de-duplication step to remove overlapping studies resulted in 375 exclusions. A further 2507 were excluded at the primary screening stage, based on their titles and/or abstracts, as they were not of relevance to the research question. These studies were excluded for reasons such as using a study sample from a small region from which generalising would not be robust, investigating other diseases, or being reviews or editorials.

During the evaluation of the full-text articles 213 were excluded for reasons such as being abstract only (52), having confounding variables (50), having incorrect population criteria (15), not measuring BMI or not defining obesity, or investigating other diseases (21), and can be found in Additional file 1: Table S6. Therefore, a total of 60 citations were included for this review (see Additional file 1: Table S7).

Of the 60 included studies, 29 reported risk equations, three reported risk of T2D, three reported the risk of a CV event, and 28 the risk of mortality. All were cohort studies or used data from cohort studies, most (49) were large (> 2000 participants) and 11 were small (< 2000 participants).

Studies reporting risk equations

Twenty-nine included studies reported an equation to calculate the risk of people with obesity or OWRC developing at least one additional complication. Of these, 12 reported equations calculating the risk of T2D [23, 29,30,31,32,33,34,35,36,37,38,39], eight of CVD [29, 38, 40,41,42,43,44,45], five of stroke [46,47,48,49,50], two of obstructive sleep apnea [51, 52], MI [49, 53], and knee osteoarthritis or replacement [54, 55], and one of heart failure [56]. Outcomes were measured and defined in various ways, with some uniformity amongst complications. For example, of studies reporting a T2D risk equation, four used clinical medical diagnosis or diabetic medication use [29, 31, 37, 38], three WHO [2] diagnostic criteria [34, 36, 39], one the American Diabetes Association criteria [33], one a UK Read Code [35], and one patient-reported [32], while one did not report a definition [23] and another used multiple definitions, including medical diagnosis and WHO diagnostic criteria [30]. Of studies reporting a CV risk equation, five used relevant International Classification of Disease codes [29, 38, 40, 42, 45], two medical records/patient surveys [41, 43], and one did not report a definition [44].

The most common risk equations reported were logistic regression models [32, 38, 39, 49, 50, 52, 55], followed by a Cox proportional hazard model [30, 36, 40, 46, 53, 54], Cox regression model [33, 42, 45, 47, 56], survival model [31, 34], QRISK2 [41, 44], and FINDRISC [23, 37] (see Table 1 for full list). Most of the equations calculated the relative risk as a hazard ratio [30, 33, 36, 40, 42, 45,46,47,48, 53, 54, 56] using a Cox proportional hazard model also referred to as a Cox regression model. Other expressions of risk included odds ratios [29, 32, 49, 50, 52, 55] (all but one of which were logistic regression models), risk scores [33, 35, 41, 43, 51], regression coefficients [29, 31], and absolute risk, e.g., percentage [23, 34, 37,38,39, 44].

Table 1 Included studies reporting equations for predicting overweight/obesity-associated comorbidities
Full size table

Risk equation building

All equations in the included studies were developed and/or validated using cohort data. These data ranged in terms of cohort size (161 [51] to 3.5 million [48] participants) and differed with regard to cohort populations used to validate the equation.

The time period over which risk was assessed ranged from 30 days [49] to lifetime [34]. The most reported (31%) risk time frame was 10 years [23, 31, 35, 37, 39, 43, 44, 47, 48], 48% of the risk equations were based on the time frame of the cohort follow-up period [30, 32, 33, 39,40,41,42, 45, 49, 52,53,54,55,56], and two did not report a time frame [46, 50].

In terms of geography, most risk equations were developed or validated with cohorts in Europe (UK [23, 30, 31, 35, 41, 42, 44, 46, 48, 52, 55], the Netherlands [29, 34, 43, 47], Norway [53, 54], Germany [50], Ireland [37], Italy [51], Russia [36], Spain [33], Sweden [56]), and the remaining on cohorts either in the USA [39, 40, 47, 49] or Australia [32, 38, 45]. One study used cohort data from two countries (the Netherlands and the USA) [47].

Validation and equation comparisons

Validation of equation performance was carried out on single or multiple data sets from internal or external cohorts. Not all studies mentioned equation validation and only six studies validated their equations with external cohorts, which is considered the most accurate validation method [23, 35, 38, 41, 43, 44]. The equations were typically tested against more than one externally validated model.

Risk predictors

As the studies measured different outcomes, there was some variation in the risk predictors included. Even when calculating the risk of developing the same complication, risk predictors varied, which could have been due to differences in equations used, individual cohort data, or definitions of the disease outcomes measured (Fig. 2, and Table 1 for full list of variables and references).

Fig. 2
figure 2

The most common risk predictors used in overweight and obesity risk equations. BMI body mass index, CVD cardiovascular disease, T2D type 2 diabetes

Full size image

In terms of population characteristics, after BMI, age was the most widely included across studies (93%). When individual background was taken into account, both deprivation (17% of studies) and education (14% of studies) were considered as well as income (3%) and family medical history (41%). Smoking was the major lifestyle risk predictor used in 83% of studies, followed by alcohol consumption (28%) and physical activity (28%). Predictors that could alleviate risk such as lifestyle intervention or pharmacotherapy were not included in any study and neither was the metabolic health of people with obesity.

Risk predictor selection

The rationale behind why risk factors were selected for the equations varied. The majority reported on their rationale. Tools such as the Framingham or QRISK2 include risk predictors that were used in six studies [23, 35, 37, 38, 41, 44] and use of previously cited risk predictors or established clinical risk factors [29, 31, 39, 48, 55] was not uncommon. Some studies designed methods to identify risk predictors specific for their study, such as Jackson et al. who considered all cohort baseline characteristics and used a backwards step regression to select the equation risk predictors [49]. A minority of studies either did not discuss risk predictor selection or their chosen variables were assessed using the cohort data at baseline as risk predictors without detailed explanation [30, 32, 34, 40, 42, 43, 45,46,47, 50,51,52, 54, 56].

BMI risk predictor

The use of BMI as a categorization of obesity levels was an inclusion criterion for this SLR, however, the range and granularity of how BMI was included into each equation differed. WHO defines weight according to six BMI ranges (underweight < 18 kg/m2, normal weight 18.5–24.9 kg/m2, overweight 25.0–29.9 kg/m2, obesity class I 30.0–34.9 kg/m2, obesity class II 35.0–39.9 kg/m2, obesity class III ≥ 40 kg/m2) yet only 13 studies used these categories [2, 29, 30, 32, 42,43,44, 48,49,50, 52,53,54,55], and only five studies included a breakdown of BMI above 30 kg/m2 [30, 42, 49, 52, 56]. Generally, most studies (18) did not report beyond BMI ≥ 30 kg/m2 in the baseline characteristics of the cohort used to develop or validate the risk equation or in the outcomes.

Quality assessment

NOS scores for the included risk equation studies ranged between eight to 14 stars with nine assessed as high quality [31, 33, 35,36,37, 39, 48, 54, 56], eight as medium [23, 29, 41, 43, 44, 47, 49, 52], and 12 as low quality [30, 32, 34, 38, 40, 42, 45, 46, 50, 51, 53, 55]. For the validation component, six out of the 29 papers scored full marks and 13 received no stars (see Additional file 1: Table S8).

Two studies achieved the highest NOS score of 14, Hippisley-Cox et al. [48] and Phillips et al. [37].

Phillips et al. [37] compared seven diabetes risk assessment tools, on the same cohort of 2047 adults, aged 50–69, with a BMI range of 28.3–32.3. Their main findings were that most diabetes risk scores offer limited ability to identify subjects with metabolic abnormalities and at risk of developing T2D. Estimates varied greatly, even when comparing risk scores that were based on the same factors (e.g. lifestyle – FINDRISC or clinical – Wilson risk scores) there was not complete concordance. The wide range of risk estimates between the risk scores were related to exclusion of certain factors, differential variable weighting, variation in high-risk thresholds, and differences in populations used to develop these scores. The study promoted the necessity to validate equations specifically for each population to identify, not just high-risk individuals, but disease hot spots that can be targeted by public health interventions.

Hippisley-Cox et al. [48] aimed to develop and validate QStroke, an algorithm to estimate risk of stroke or transient ischemic attack, in 3.5 million adults aged 25–84 years old. They compared QStroke to CHADS2 and CHA2DS2VASc in patients with atrial fibrillation, and they compared it to the Framingham stroke score in the stroke/transient ischemic attack free population. QStroke had improved performance compared against all three scores although levels of discrimination were lower for patients with atrial fibrillation. In atrial fibrillation patients there is a slight over prediction for women at high levels of predicted risk, whereas the model is well calibrated for men. They identified a specific population, those for whom anticoagulation is considered, that could particularly benefit from Qstroke. Qstroke incorporates weighted established risk factors, such as ethnicity and deprivation, that are absent from existing stroke risk tools. It is also more patient friendly, in terms of giving absolute measures of stroke risk and it is aimed at being incorporated into GP practices.

Studies reporting absolute or proportional risk

Risk of T2D

Three studies reported the risk of developing T2D in a population with obesity (BMI ≥ 30 kg/m2) and NGT or prediabetes [34, 57, 58] and were carried out in the Netherlands [34], Poland [58], or the USA [57] (Table 2).

Table 2 Included studies reporting T2D risk in a population with obesity with NGT or prediabetes
Full size table

Only one study split obesity into multiple categories, BMI 30–34.9 kg/m2, 35–39.9 kg/m2 including a separate one for ≥ 40 kg/m2 (class III obesity) [57], the others grouped obesity classes together, including the Zatonska et al. study that used only one category ≥ 30 kg/m2 [58]. All studies reported an increased risk of developing T2D, either with absolute risk [34] or relative risk [57, 58].

On the NOS quality assessment scale, two of the studies were of low quality (10 stars) [34, 57] and one of medium quality (11 stars) [58] (see Additional file 1: Table S8).

Risk of CV event

Three studies reported the risk of experiencing a CV event in a population with class III obesity and NGT or T2D [59,60,61] and were carried out in Norway [60], Sweden [59], and the UK [61] (Table 3).

Table 3 Included studies reporting the risk of CV events in populations with severe obesity with NGT/T2D
Full size table

All studies reported that class III obesity increased the relative risk of a CV event, except in the case of metabolically healthy obesity (obesity without hypertension, dyslipidaemia, and insulin resistance) where there was no increased risk of MI [60].

Two of the studies were of medium quality (11 stars) [59, 60] and one of high quality (13 stars) [61] according to the NOS rating (see Additional file 1: Table S8).

Risk of mortality

Three studies reported the risk of mortality from total knee arthroplasty [62,63,64], four from stroke [65,66,67,68], and 22 from CV events in a population with obesity or OWRC [53, 61, 67, 69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87] (Table 4). Studies were carried out in Australia (one study), Europe (14 studies over five countries), and the USA (13 studies).

Table 4 Included studies reporting mortality risk from CV events/total knee arthroplasty/stroke in populations with obesity or OWRC
Full size table

Almost all studies measured relative risk. Most studies reported risk of mortality increased with higher BMI counts, however, not all studies agreed. Two studies reporting on mortality following total knee arthroplasty showed no association with weight [63, 64]. The often-documented paradoxical link with mortality following stroke was reported in two of three studies [65, 66]. For CV linked mortality, a paradoxical link was reported in two studies [69, 72], two reported no association [78, 85], and in others a more nuanced association was demonstrated [53, 61, 70, 76, 77, 83].

The NOS rating for the studies on mortality from total knee arthroplasty were all of high quality. Of studies detailing mortality from stroke two were assessed as low quality (10 stars) [66, 67], one medium (11 stars) [68], and one high (13 stars) [65]. The majority (12) of studies reporting on risk of mortality from CV events were of medium quality [69,70,71, 73, 77, 81,82,83,84,85,86,87], there were seven of low [53, 67, 72, 74,75,76, 80], and three of high quality [61, 78, 79].One study achieved the maximum star in this SLR of 14, Das et al. [79]. They assessed the impact of class III obesity on care and outcomes on a specific type of MI, ST-segment MI, in a cohort of 49,329 subjects. Class III obesity was associated with younger age at MI presentation, as well as a higher prevalence of diabetes, hypertension, and dyslipidaemia but less extensive coronary artery disease and better left ventricular systolic function, though in-hospital mortality remains significantly higher than those with class I obesity. Patients with class III obesity and MI were more than a decade younger than normal weight counterparts and more likely to be women and of African American descent.

Discussion

This SLR represents the current state on available data for risk equations that can predict the incidence of complications in populations with OWRC and obesity. It further exposes the unmet need for obesity specific equations, the limited ability of these equations to estimate risk for higher BMI categories, and the inconsistency of results between different risk equations on the same populations with OWRC and obesity. There is a clear necessity for new data or adjustment of the current risk equations to allow for more robust health economic modelling that can inform accurate decision-making strategies for the management of people living with obesity and specifically class III obesity.

It is important to validate risk equations in external cohorts to assess their usability in different populations and regions [37]. Otherwise potential problems arise when developing multivariable risk prediction models, including overestimating performance and overfitting [6, 21], meaning that the equation’s performance cannot be applied to external data cohorts not used to derive the model [88]. Beyond this, Mathur et al. acknowledges that even though many risk prediction tools have been developed, for T2D for example, relatively few validated risk scoring systems have demonstrated real-world application [35].

Only six studies reported on external validation of the risk equations [23, 35, 38, 41, 43, 44] (eight out of the total 60 included studies used external validation [23, 35, 38, 41, 43, 44, 79, 84]). All of the externally validated equations calculate the risk of T2D and CVD, but none calculated the risk of developing other complications commonly associated with obesity or overweight. However, even when comparing the performance of frequently used and validated risk equations in T2D and CVD, Gray et al. (2014 and 2015) demonstrated that they produce different outcomes and assessments of risk [23, 44]. This highlights the complex nature of these equations and the impact of the specific risk predictors and how they are calibrated.

A common theme in the literature is the lack of granularity among BMI classes of obesity used to develop or validate the published risk equations. People living with obesity were typically characterized as having a BMI of ≥ 30 kg/m2 and not broken down further. Therefore, it can be assumed that these risk equations may work well up to a certain BMI level before they lose sensitivity. Only 17% of the risk equations detailed the BMI of the cohort used to develop and/or validate the risk equation by obesity class beyond ≥ 30 kg/m2 [30, 42, 49, 52, 56]. Most (62%) risk equation studies did not report any BMI breakdown of the cohort used or only reported the average BMI at best [23, 29, 31,32,33,34,35, 37, 39,40,41, 44, 47, 48, 50, 51, 54, 55]. No risk equation was developed or validated using a cohort separately distinguished beyond BMI ≥ 40 kg/m2. Even with the lack of risk equation specificity for class III obesity some still calculated the risk of developing a complication more granularly, beyond this point. No explanation or insight was identified for including higher BMI groups into these risk equations, or how, if at all, a presumably small sample within the highest BMI groups was dealt with accurately to create these equations. The representation of each population in the development and validation process lacked clarity. Therefore, the accuracy of the calibration for calculating class III obesity specific risk and the subsequent performance of the risk equation is contentious.

Absolute or relative risks, as identified in this SLR, can be used in calibration of risk equations to improve accuracy in higher BMI populations, or in populations not well represented (e.g., those with specific complications). Poorly calibrated risk equations can be misleading and potentially harmful in clinical decisions. The identified calibration targets can either be used to calibrate an existing model to better fit populations with obesity or overweight with more precision, or to confirm if new risk equations are robust. Such risk measures can be used as calibration targets in simulations to help optimize risk equation parameters, for example [89]. This approach would ensure greater accuracy and generalizability in subpopulations currently not well represented.

As for the identified studies reporting absolute or proportional risk for development of T2D, CV events, and mortality in specific populations with obesity or OWRC, those assessing outcome and complication development often failed to distinguish multiple obesity classes and lacked precision. The results of different studies on the same topic were not always in agreement, making it difficult to draw concrete and meaningful conclusions. For example, different studies reported higher, lower, and no association for stroke mortality risk in association with obesity and OWRC [65,66,67,68]. Also, for CV mortality risk in association with obesity and OWRC some studies have reported a paradoxical link [69, 72] or no association [78, 85], countering increased [67, 74, 79,80,81, 84, 86, 87] or nuanced risk [53, 61, 70, 76, 77, 83] from other studies. More consistency in studies is needed, with more granulated risk predictors (e.g., BMI) as well as a wider range of risk factors being explored, such as historical and ongoing weight increases rather than just the current BMI. This would allow robust outcomes to inform calibration and validation for risk equations.

The choice of risk predictor usage and its weighting depend on the outcome and population of interest. There is not one catch all risk predictor for weight related complications. If we consider patients with T2D, BMI is a strong independent risk factor of heart failure in a stepwise fashion, but not for acute myocardial infarction [49]. In addition, risk predictor choice may be limited by availability. For example, in primary care settings potential risk predictors such as liver function tests, pulse rate, and BMI are not always recorded [31], so what makes a good risk predictor also lies in its ease of use. In the absence of singular strong risk predictors the use of multiple risk predictors that individually slightly increase risk can still identify a high risk group [29].

While not perfect, risk equations have been used to inform both clinical and health technology assessment (HTA) decision making in obesity and overweight [90]. The UK’s NICE guidelines recommend using the Cambridge Risk Score, FINDRISC, Leicester Risk Assessment, and QDiabetes for calculating the risk of developing T2D [91]. In a comparison of the four, Gray et al. concluded that either QDiabetes or FINDRISC are the most practical and cost-effective tools for NICE decision making [23]. Mathur et al. concluded that NICE could use the QDScore to identify high T2D risk people, however this SLR did not identify any other studies comparing this risk equation with other tools and therefore this may not be substantiated [35]. In CVD, NICE recommend the use of QRISK2 because Framingham equations overestimate the risk in a UK population and QRISK2 is widely used in UK general practice [41, 48]. Indeed, when compared with three other validated CVD risk equations, QRISK2 was reported to have high accuracy and is frequently updated, making it most appropriate for decision making [44].

Conclusion

The studies included in this SLR used risk equations adapted for the target population but not built specifically for them and BMI categories that were inconsistently reported and adhered to, providing limited potential for comparison or accurate risk assessment. The included studies calculating absolute or proportional risk of specific disease outcomes can be built upon to inform more robust risk equations and so, provide more reliable health economic models to aid decision makers.

The accuracy of the risk equations when applied to other populations and between BMI classes, in particular class III obesity, remains questionable. The majority of risk equations lacked external validation and those that did only calculated the risk of developing T2D and CVD. The risk equations were largely limited to European cohorts and made little attempt to delineate between obesity classes, particularly with regard to higher BMI groups.

This SLR also identified a number of studies reporting absolute risks of developing T2D, CVD or mortality from CVD, T2D, or total knee arthroplasty in populations with obesity and OWRC. Such absolute scores can be used as calibration targets when calibrating risk equations, to help with accuracy in populations with higher BMIs or in subgroups of populations with obesity or OWRC.

Implications in practice and recommendations

This SLR has highlighted several gaps to be addressed in future work, with real-world implications. Further work is needed to externally validate equations calculating the risk of populations with obesity or OWRC developing a complication. This should also include validating these equations in diverse populations to ensure generalizability. It should additionally ensure cohorts with obesity are clearly broken down by class, especially focusing on class III obesity, again to ensure accuracy in this population. We identified externally validated risk equations used to predict T2D and CV risk but, particularly for other obesity associated complications, found limited data and further development for all outcomes is required. Other approaches could include utilizing or generating real-world evidence to fill some of the data gaps and recalibrating existing risk equations. This will help to inform robust evaluations of obesity and overweight treatments, HTA decision making, and beyond. More studies informing on the absolute or proportional risk of a wider range of risk factors, specifically for populations ranging from overweight through class I, II, and III obesity, would help to calibrate and improve both existing and new risk equations for these populations.

Many of the points raised above can enhance future risk equations and study design. We have discussed in detail the importance of categorising BMI and recommend that future studies harmonise through using, for example, the relevant WHO BMI scales, and avoid grouping weight categories. In addition to identifying specific risk, it will allow individual data sets to be pooled and reanalysed. Additional guidelines could include new risk equations focusing on weight associated complications that are poorly predicted by the existing risk equations such as metabolic abnormalities and risk of developing T2D. Also, the identification of new risk predictors through methods such as genetic profiling and related to metabolic alterations would add value, while new data can support existing risk predictors and determine how they are best weighted for each equation.

As scientists increasingly understand the value of data sharing and open-source codes, big data will further change the way we predict risk. As an example, all modern predictive models of financial market risks are based on data driven analysis exploiting machine learning. One of the main reasons this is not done in medicine yet is due to the data deficit needed for a multifactorial analysis. Only a few areas in the medical field, such as radiomics, have harnessed this power with data driven analysis for diagnosis and prognosis of several diseases. As more data is gathered, risk modelling for weight related outcomes will become more powerful to accurately predict outcomes for populations living with obesity, populations that are growing rapidly in many parts of the world.

Limitations

This SLR was limited to countries by geographical scope (Australia, Canada, Europe, and the USA) and therefore may have excluded potentially useful risk equations developed in other regions. Results predominantly reflected European studies, while Canada and the USA are not well represented. Furthermore, because a direct comparison of the risk equations was not possible due to each being developed or calculated using different methods for different purposes, a judgment on how each risk score performs is the opinion of the user or decision-maker in a particular context. Finally, obesity may not be a single homogenous disease. Thus, if there are different obesity subtypes leading to the same BMI in different people then a single unifying risk equation may not be possible, but rather better diagnostic ability of these obesity subtypes may be required to refine future risk equations.

Availability of data and materials

The following electronic databases were searched as standard evidence sources: MEDLINE, MEDLINE In-Process and e-publications ahead-of-print (via PubMed), Embase (via Embase.com), the technology appraisal submissions to the National Institute for Clinical Excellence (NICE) using risk equations, and bibliographic reference lists from included studies. Search strings can be found in Additional file 1: Table S2 (Embase) and Additional file 1: Table S3 (PubMed).

Abbreviations

ACE:

Angiotensin converting enzyme

AR:

Absolute risk

BMI:

Body mass index

CI:

Confidence interval

CV:

Cardiovascular

CVD:

Cardiovascular disease

FINDRISC:

Finnish Diabetes Risk Score

HbA1c :

Glycated hemoglobin

HR:

Hazard ratio

HTA:

Health technology assessment

MI:

Myocardial infarction

NGT:

Normal glucose tolerance

NICE:

National Institute for Clinical Excellence

NOS:

Newcastle Ottawa Scale

OR:

Odds ratio

OWRC:

Overweight with a weight-related complication

PRISMA:

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

SLR:

Systematic literature review

STEMI:

ST-elevation myocardial infarction

T2D:

Type 2 diabetes

WHO:

World Health Organization

References

  1. Abdelaal M, le Roux CW, Docherty NG. Morbidity and mortality associated with obesity. Ann Transl Med. 2017;5(7):161.

    Article  Google Scholar 

  2. World Health Organization. Obesity and overweight. 2020. https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight. Accessed 1 June 2021.

  3. Purnell JQ. Definitions, classification, and epidemiology of obesity. Endotext [Internet]; 2018.

  4. Flegal KM, Graubard BI. Estimates of excess deaths associated with body mass index and other anthropometric variables. Am J Clin Nutr. 2009;89(4):1213–9.

    Article  CAS  Google Scholar 

  5. Willett K, Jiang R, Lenart E, Spiegelman D, Willett W. Comparison of bioelectrical impedance and BMI in predicting obesity-related medical conditions. Obesity (Silver Spring). 2006;14(3):480–90.

    Article  Google Scholar 

  6. Padula WV, Allen RR, Nair KV. Determining the cost of obesity and its common comorbidities from a commercial claims database. Clin Obes. 2014;4(1):53–8.

    Article  CAS  Google Scholar 

  7. Andreyeva T, Sturm R, Ringel JS. Moderate and severe obesity have large differences in health care costs. Obes Res. 2004;12(12):1936–43.

    Article  Google Scholar 

  8. Wang YC, Pamplin J, Long MW, Ward ZJ, Gortmaker SL, Andreyeva T. Severe obesity in adults cost state Medicaid programs nearly $8 billion in 2013. Health Aff (Millwood). 2015;34(11):1923–31.

    Article  Google Scholar 

  9. Ward ZJ, Bleich SN, Cradock AL, Barrett JL, Giles CM, Flax C, et al. Projected U.S. state-level prevalence of adult obesity and severe obesity. N Engl J Med. 2019;381(25):2440–50.

    Article  Google Scholar 

  10. Wang YC, McPherson K, Marsh T, Gortmaker SL, Brown M. Health and economic burden of the projected obesity trends in the USA and the UK. Lancet. 2011;378(9793):815–25.

    Article  Google Scholar 

  11. De Lorenzo A, Gratteri S, Gualtieri P, Cammarano A, Bertucci P, Di Renzo L. Why primary obesity is a disease? J Transl Med. 2019;17(1):169.

    Article  Google Scholar 

  12. Sinaga M, Worku M, Yemane T, Tegene E, Wakayo T, Girma T, et al. Optimal cut-off for obesity and markers of metabolic syndrome for Ethiopian adults. Nutr J. 2018;17(1):109.

    Article  Google Scholar 

  13. WHO Expert Consultation. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet. 2004;363(9403):157–63.

    Article  Google Scholar 

  14. Misra A, Chowbey P, Makkar BM, Vikram NK, Wasir JS, Chadha D, et al. Consensus statement for diagnosis of obesity, abdominal obesity and the metabolic syndrome for Asian Indians and recommendations for physical activity, medical and surgical management. J Assoc Physicians India. 2009;57:163–70.

    CAS  Google Scholar 

  15. Kitahara CM, Flint AJ, Berrington de Gonzalez A, Bernstein L, Brotzman M, MacInnis RJ, et al. Association between class III obesity (BMI of 40–59 kg/m2) and mortality: a pooled analysis of 20 prospective studies. PLoS Med. 2014;11(7):e1001673.

    Article  Google Scholar 

  16. Kontis V, Mathers CD, Rehm J, Stevens GA, Shield KD, Bonita R, et al. Contribution of six risk factors to achieving the 25×25 non-communicable disease mortality reduction target: a modelling study. Lancet. 2014;384(9941):427–37.

    Article  Google Scholar 

  17. World Health Organization. Global action plan for the prevention and control of noncommunicable diseases 2013–2020. 2013. https://apps.who.int/iris/bitstream/handle/10665/94384/?sequence=1. Accessed 30 June 2021.

  18. Kluge HHP, Wickramasinghe K, Rippin HL, Mendes R, Peters DH, Kontsevaya A, et al. Prevention and control of non-communicable diseases in the COVID-19 response. Lancet. 2020;395(10238):1678–80.

    Article  CAS  Google Scholar 

  19. World Health Organization. Noncommunicable diseases in emergencies. 2016. https://apps.who.int/iris/rest/bitstreams/908919/retrieve. Accessed 30 June 2021

  20. Li Y, Sperrin M, Belmonte M, Pate A, Ashcroft DM, van Staa TP. Do population-level risk prediction models that use routinely collected health data reliably predict individual risks? Sci Rep. 2019;9(1):11222.

    Article  Google Scholar 

  21. Collins G, Mallett S, Omar O, Yu LM. Developing risk prediction models for type 2 diabetes: a systematic review of methodology and reporting. BMC Med. 2011;9(103):1–14.

    Google Scholar 

  22. Pate A, Emsley R, Ashcroft DM, Brown B, van Staa T. The uncertainty with using risk prediction models for individual decision making: an exemplar cohort study examining the prediction of cardiovascular disease in English primary care. BMC Med. 2019;17(1):134.

    Article  Google Scholar 

  23. Gray BJ, Bracken RM, Turner D, Morgan K, Thomas M, Williams SP, et al. Different type 2 diabetes risk assessments predict dissimilar numbers at ‘high risk’: a retrospective analysis of diabetes risk-assessment tools. Br J Gen Pract. 2015;65(641):e852–60.

    Article  Google Scholar 

  24. van Staa TP, Gulliford M, Ng ES, Goldacre B, Smeeth L. Prediction of cardiovascular risk using Framingham, ASSIGN and QRISK2: how well do they predict individual rather than population risk? PLoS ONE. 2014;9(10):e106455.

    Article  Google Scholar 

  25. Tacconelli E. Systematic reviews: CRD’s guidance for undertaking reviews in health care. Lancet Infect Dis. 2010;10(4):226.

    Article  Google Scholar 

  26. Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009;339:b2535.

    Article  Google Scholar 

  27. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71.

    Article  Google Scholar 

  28. Wells G, Shea B, O’Connell D, Peterson J, Welch V, Losos M, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2009. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp. Accessed 30 June 2021.

  29. Alssema M, Newson R, Bakker S, Stehouwer C, Heymans M, Nijpels G, et al. One risk assessment tool for cardiovascular disease, type 2 diabetes, and chronic kidney disease. Diabetes Care. 2012;35(4):741–8.

    Article  Google Scholar 

  30. Booth H, Khan O, Prevost T, Reddy M, Dregan A, Charlton J, et al. Incidence of type 2 diabetes after bariatric surgery: population-based matched cohort study. Lancet Diabetes Endocrinol. 2014;2(12):963–8.

    Article  Google Scholar 

  31. Coles B, Khunti K, Booth S, Zaccardi F, Davies MJ, Gray LJ. Prediction of type 2 diabetes risk in people with non-diabetic hyperglycaemia: model derivation and validation using UK primary care data. BMJ Open. 2020;10(10):e037937.

    Article  Google Scholar 

  32. Ding D, Chong S, Jalaludin B, Comino E, Bauman AE. Risk factors of incident type 2-diabetes mellitus over a 3-year follow-up: results from a large Australian sample. Diabetes Res Clin Pract. 2015;108(2):306–15.

    Article  Google Scholar 

  33. Guasch-Ferre M, Bullo M, Costa B, Martinez-Gonzalez MA, Ibarrola-Jurado N, Estruch R, et al. A risk score to predict type 2 diabetes mellitus in an elderly Spanish Mediterranean population at high cardiovascular risk. PLoS ONE. 2012;7(3):e33437.

    Article  CAS  Google Scholar 

  34. Ligthart S, van Herpt TTW, Leening MJG, Kavousi M, Hofman A, Stricker BHC, et al. Lifetime risk of developing impaired glucose metabolism and eventual progression from prediabetes to type 2 diabetes: a prospective cohort study. Lancet Diabetes Endocrinol. 2016;4(1):44–51.

    Article  CAS  Google Scholar 

  35. Mathur R, Noble D, Smith D, Greenhalgh T, Robson J. Quantifying the risk of type 2 diabetes in East London using the QDScore: a cross-sectional analysis. Br J Gen Pract. 2012;62(603):e663–70.

    Article  Google Scholar 

  36. Mustafina SV, Rymar OD, Shcherbakova LV, Verevkin EG, Pikhart H, Sazonova OV, et al. The risk of type 2 diabetes mellitus in a Russian population cohort according to data from the HAPIEE project. J Pers Med. 2021;11(2):119.

    Article  Google Scholar 

  37. Phillips CM, Kearney PM, McCarthy VJ, Harrington JM, Fitzgerald AP, Perry IJ. Comparison of diabetes risk score estimates and cardiometabolic risk profiles in a middle-aged Irish population. PLoS ONE. 2013;8(11):e78950.

    Article  Google Scholar 

  38. Rauh SP, Rutters F, van der Heijden A, Luimes T, Alssema M, Heymans MW, et al. External validation of a tool predicting 7-year risk of developing cardiovascular disease, type 2 diabetes or chronic kidney disease. J Gen Intern Med. 2018;33(2):182–8.

    Article  Google Scholar 

  39. Wilkinson L, Yi N, Mehta T, Judd S, Garvey WT. Development and validation of a model for predicting incident type 2 diabetes using quantitative clinical data and a Bayesian logistic model: a nationwide cohort and modeling study. PLoS Med. 2020;17(8):e1003232.

    Article  CAS  Google Scholar 

  40. Bruce SA. The association between central fat distribution and recurrent cardiovascular disease events in female survivors of nonfatal myocardial infarction. J Cardiovasc Nurs. 2015;30(2):E15–22.

    Article  Google Scholar 

  41. Chang KCM, Lee JT, Vamos EP, Soljak M, Johnston D, Khunti K, et al. Impact of the National Health Service Health Check on cardiovascular disease risk: a difference-in-differences matching analysis. CMAJ. 2016;188(10):E228–38.

    Article  Google Scholar 

  42. Costanzo P, Cleland JG, Pellicori P, Clark AL, Hepburn D, Kilpatrick ES, et al. The obesity paradox in type 2 diabetes mellitus: relationship of body mass index to prognosis: a cohort study. Ann Intern Med. 2015;162(9):610–8.

    Article  Google Scholar 

  43. de Boer AW, de Mutsert R, den Heijer M, Jukema JW, Rosendaal FR, Blom JW, et al. Overweight can be used as a tool to guide case-finding for cardiovascular risk assessment. Fam Pract. 2015;32(6):646–51.

    Article  Google Scholar 

  44. Gray BJ, Bracken RM, Turner D, Morgan K, Mellalieu SD, Thomas M, et al. Predicted 10-year risk of cardiovascular disease is influenced by the risk equation adopted: a cross-sectional analysis. Br J Gen Pract. 2014;64(627):e634–40.

    Article  Google Scholar 

  45. Joshy G, Korda RJ, Attia J, Liu B, Bauman AE, Banks E. Body mass index and incident hospitalisation for cardiovascular disease in 158 546 participants from the 45 and Up Study. Int J Obes. 2014;38(6):848–56.

    Article  CAS  Google Scholar 

  46. Burns DJP, Rapetto F, Angelini GD, Benedetto U, Caputo M, Ciulli F, et al. Body mass index and early outcomes following mitral valve surgery for degenerative disease. J Thorac Cardiovasc Surg. 2021;161(5):1765-1773.e2.

    Article  Google Scholar 

  47. Ferket BS, Van Kempen BJH, Wieberdink RG, Steyerberg EW, Koudstaal PJ, Hofman A, et al. Separate prediction of intracerebral hemorrhage and ischemic stroke. Neurology. 2014;82(20):1804–12.

    Article  Google Scholar 

  48. Hippisley-Cox J, Coupland C, Brindle P. Derivation and validation of QStroke score for predicting risk of ischaemic stroke in primary care and comparison with other risk scores: a prospective open cohort study. BMJ. 2013;346:f2573.

    Article  Google Scholar 

  49. Jackson RS, Black JH III, Lum YW, Schneider EB, Freischlag JA, Perler BA, et al. Class I obesity is paradoxically associated with decreased risk of postoperative stroke after carotid endarterectomy. J Vasc Surg. 2012;55(5):1306–12.

    Article  Google Scholar 

  50. Winter Y, Pieper L, Klotsche J, Riedel O, Wittchen HU. Obesity and abdominal fat markers in patients with a history of stroke and transient ischemic attacks. J Stroke Cerebrovasc Dis. 2016;25(5):1141–7.

    Article  Google Scholar 

  51. Ahlin S, Manco M, Panunzi S, Verrastro O, Giannetti G, Prete A, et al. A new sensitive and accurate model to predict moderate to severe obstructive sleep apnea in patients with obesity. Medicine (Baltimore). 2019;98(32):e16687.

    Article  Google Scholar 

  52. Erridge S, Moussa O, McIntyre C, Hariri A, Tolley N, Kotecha B, et al. Obstructive sleep apnea in obese patients: a UK population analysis. Obes Surg. 2021;31(5):1986–93.

    Article  CAS  Google Scholar 

  53. Borgeraas H, Hertel JK, Svingen GFT, Seifert R, Pedersen EKR, Schartum-Hansen H, et al. Association of body mass index with risk of acute myocardial infarction and mortality in Norwegian male and female patients with suspected stable angina pectoris: a prospective cohort study. BMC Cardiovasc Disord. 2014;14:68.

    Google Scholar 

  54. Apold H, Meyer H, Nordsletten L, Furnes O, Baste V, Flugsrud GB. Risk factors for knee replacement due to primary osteoarthritis, a population based, prospective cohort study of 315,495 individuals. BMC Musculoskelet Disord. 2014;15(217):1–11.

    Google Scholar 

  55. Zhang W, McWilliams DF, Ingham SL, Doherty SA, Muthuri S, Muir KR, et al. Nottingham knee osteoarthritis risk prediction models. Ann Rheum Dis. 2011;70(9):1599–604.

    Article  Google Scholar 

  56. Glogner S, Rosengren A, Olsson M, Gudbjörnsdottir S, Svensson AM, Lind M. The association between BMI and hospitalization for heart failure in 83,021 persons with type 2 diabetes: a population-based study from the Swedish National Diabetes Registry. Diabet Med. 2014;31(5):586–94.

    Article  CAS  Google Scholar 

  57. Blume SW, Li Q, Huang JC, Hammer M, Graf TR. Variation in the risk of progression between glycemic stages across different levels of body mass index: evidence from a United States electronic health records system. Curr Med Res Opin. 2015;31(1):115–24.

    Article  CAS  Google Scholar 

  58. Zatońska K, Basiak-Rasała A, Różańska D, Karczewski M, Wołyniec M, Szuba A, et al. Changes in diabetes prevalence and corresponding risk factors - findings from 3- and 6-year follow-up of PURE Poland cohort study. BMC Public Health. 2020;20(1):843.

    Article  Google Scholar 

  59. Edqvist J, Rawshani A, Adiels M, Björck L, Lind M, Svensson AM, et al. Contrasting associations of body mass index and hemoglobin A1c on the excess risk of acute myocardial infarction and heart failure in type 2 diabetes mellitus. J Am Heart Assoc. 2019;8(24):e013871.

    Article  Google Scholar 

  60. Mørkedal B, Vatten LJ, Romundstad PR, Laugsand LE, Janszky I. Risk of myocardial infarction and heart failure among metabolically healthy but obese individuals: HUNT (Nord-Trondelag Health Study), Norway. J Am Coll Cardiol. 2014;63(11):1071–8.

    Article  Google Scholar 

  61. Iyen B, Weng S, Vinogradova Y, Akyea RK, Qureshi N, Kai J. Long-term body mass index changes in overweight and obese adults and the risk of heart failure, cardiovascular disease and mortality: a cohort study of over 260,000 adults in the UK. BMC Public Health. 2021;21(1):576.

    Article  Google Scholar 

  62. Alvi HM, Mednick RE, Krishnan V, Kwasny MJ, Beal MD, Manning DW. The effect of BMI on 30 day outcomes following total joint arthroplasty. J Arthroplasty. 2015;30(7):1113–7.

    Article  Google Scholar 

  63. Sing DC, Luan TF, Feeley BT, Zhang AL. Is obesity a risk factor for adverse events after knee arthroscopy? Arthroscopy. 2016;32(7):1346-1353.e1.

    Article  Google Scholar 

  64. Sundaram K, Warren J, Anis H, George J, Murray T, Higuera CA, et al. An increased body mass index was not associated with higher rates of 30-day postoperative complications after unicompartmental knee arthroplasty. Knee. 2019;26(3):720–8.

    Article  Google Scholar 

  65. Andersen KK, Olsen TS. The obesity paradox in stroke: lower mortality and lower risk of readmission for recurrent stroke in obese stroke patients. Int J Stroke. 2015;10(1):99–104.

    Article  Google Scholar 

  66. Hoffman H, Jalal MS, Furst T, Chin LS. The obesity paradox in spontaneous intracerebral hemorrhage: results from a retrospective analysis of the nationwide inpatient sample. Neurocrit Care. 2020;32(3):765–74.

    Article  Google Scholar 

  67. Silventoinen K, Tynelius P, Rasmussen F. Weight status in young adulthood and survival after cardiovascular diseases and cancer. Int J Epidemiol. 2014;43(4):1197–204.

    Article  Google Scholar 

  68. Skolarus LE, Sanchez BN, Levine DA, Baek J, Kerber KA, Morgenstern LB, et al. Association of body mass index and mortality after acute ischemic stroke. Circ Cardiovasc Qual Outcomes. 2014;7(1):64–9.

    Article  Google Scholar 

  69. Batsis JA, Singh S, Lopez-Jimenez F. Anthropometric measurements and survival in older Americans: results from the third National Health and Nutrition Examination Survey. J Nutr Health Aging. 2014;18(2):123–30.

    Article  CAS  Google Scholar 

  70. Perotto M, Panero F, Gruden G, Fornengo P, Lorenzati B, Barutta F, et al. Obesity is associated with lower mortality risk in elderly diabetic subjects: the Casale Monferrato study. Acta Diabetol. 2013;50(4):563–8.

    Article  CAS  Google Scholar 

  71. Crotti G, Gianfagna F, Bonaccio M, Di Castelnuovo A, Costanzo S, Persichillo M, et al. Body mass index and mortality in elderly subjects from the Moli-sani Study: a possible mediation by low-grade inflammation? Immunol Invest. 2018;47(8):774–89.

    Article  CAS  Google Scholar 

  72. Ratwatte S, Hyun K, D’Souza M, Barraclough J, Chew DP, Shetty P, et al. Relation of body mass index to outcomes in acute coronary syndrome. Am J Cardiol. 2021;138:11–9.

    Article  Google Scholar 

  73. Bhaskaran K, dos-Santos-Silva I, Leon DA, Douglas IJ, Smeeth L. Association of BMI with overall and cause-specific mortality: a population-based cohort study of 3·6 million adults in the UK. Lancet Diabetes Endocrinol. 2018;6(12):944–53.

    Article  Google Scholar 

  74. Boggs DA, Rosenberg L, Cozier YC, Wise LA, Coogan PF, Ruiz-Narvaez EA, et al. General and abdominal obesity and risk of death among black women. N Engl J Med. 2011;365(10):901–8.

    Article  CAS  Google Scholar 

  75. Champagne-Langabeer T, Kim J, Bower JK, Gardner A, Fowler R, Langabeer JR II. Obesity, treatment times, and cardiovascular outcomes after ST-elevation myocardial infarction: findings from Mission: Lifeline North Texas. J Am Heart Assoc. 2017;6(9):e005827.

    Article  Google Scholar 

  76. Bo S, Musso G, Gambino R, Villois P, Gentile L, Durazzo M, et al. Prognostic implications for insulin-sensitive and insulin-resistant normal-weight and obese individuals from a population-based cohort. Am J Clin Nutr. 2012;96(5):962–9.

    Article  CAS  Google Scholar 

  77. Calori G, Lattuada G, Piemonti L, Garancini MP, Ragogna F, Villa M, et al. Prevalence, metabolic features, and prognosis of metabolically healthy obese Italian individuals: the Cremona study. Diabetes Care. 2011;34(1):210–5.

    Article  Google Scholar 

  78. Czernichow S, Kengne AP, Stamatakis E, Hamer M, Batty GD. Body mass index, waist circumference and waist-hip ratio: which is the better discriminator of cardiovascular disease mortality risk?: evidence from an individual-participant meta-analysis of 82 864 participants from nine cohort studies. Obes Rev. 2011;12(9):680–7.

    CAS  Google Scholar 

  79. Das SR, Alexander KP, Chen AY, Powell-Wiley TM, Diercks DB, Peterson ED, et al. Impact of body weight and extreme obesity on the presentation, treatment, and in-hospital outcomes of 50,149 patients with ST-segment elevation myocardial infarction: results from the NCDR (National Cardiovascular Data Registry). J Am Coll Cardiol. 2011;58(25):2642–50.

    Article  Google Scholar 

  80. Eeg-Olofsson K, Gudbjörnsdottir S, Eliasson B, Zethelius B, Cederholm J, NDR. The triglycerides-to-HDL-cholesterol ratio and cardiovascular disease risk in obese patients with type 2 diabetes: an observational study from the Swedish National Diabetes Register (NDR). Diabetes Res Clin Pract. 2014;106(1):136–44.

    Article  CAS  Google Scholar 

  81. Ma J, Flanders WD, Ward EM, Jemal A. Body mass index in young adulthood and premature death: analyses of the US National Health Interview Survey linked mortality files. Am J Epidemiol. 2011;174(8):934–44.

    Article  Google Scholar 

  82. Ma J, Jemal A, Flanders WD, Ward EM. Joint association of adiposity and smoking with mortality among U.S. adults. Prev Med. 2013;56(3–4):178–84.

    Article  Google Scholar 

  83. McAuley PA, Artero EG, Sui X, Lavie CJ, Almeida MJ, Blair SN. Fitness, fatness, and survival in adults with prediabetes. Diabetes Care. 2014;37(2):529–36.

    Article  CAS  Google Scholar 

  84. Payvar S, Kim S, Rao SV, Krone R, Neely M, Paladugu N, et al. In-hospital outcomes of percutaneous coronary interventions in extremely obese and normal-weight patients: findings from the NCDR (National Cardiovascular Data Registry). J Am Coll Cardiol. 2013;62(8):692–6.

    Article  Google Scholar 

  85. Hotchkiss JW, Davies CA, Leyland AH. Adiposity has differing associations with incident coronary heart disease and mortality in the Scottish population: cross-sectional surveys with follow-up. Int J Obes (Lond). 2013;37(5):732–9.

    Article  CAS  Google Scholar 

  86. Kjøllesdal M, Degerud E, Næss O. The association between obesity and cardiovascular disease mortality in different strata of socioeconomic position: evidence from pooled Norwegian health surveys. Eur J Public Health. 2019;29(6):1160–6.

    Article  Google Scholar 

  87. Kjøllesdal MKR, Smith GD, Ariansen I, Kinge JM, Degerud E, Næss O. The association between BMI and mortality using early adulthood BMI as an instrumental variable for midlife BMI. Sci Rep. 2018;8(1):11499.

    Article  Google Scholar 

  88. Babyak MA. What you see may not be what you get: a brief, nontechnical introduction to overfitting in regression-type models. Psychosom Med. 2004;66(3):411–21.

    Google Scholar 

  89. Vanni T, Karnon J, Madan J, White RG, Edmunds WJ, Foss AM, et al. Calibrating models in economic evaluation: a seven-step approach. Pharmacoeconomics. 2011;29(1):35–49.

    Article  Google Scholar 

  90. Basu S, Sussman JB, Berkowitz SA, Hayward RA, Bertoni AG, Correa A, et al. Validation of Risk Equations for Complications of Type 2 Diabetes (RECODe) using individual participant data from diverse longitudinal cohorts in the U.S. Diabetes Care. 2018;41(3):586–95.

    Article  Google Scholar 

  91. National Institute for Health and Care Excellence. Type 2 diabetes: prevention in people at high risk. PH 38. 2012. https://www.nice.org.uk/guidance/ph38. Accessed 01 July 2021.

Download references

Acknowledgements

Protocol development, running of searches, and hand searching of medical societies were supported by Katherine MacGilchrist, Epidemica, Oxfordshire, UK. Writing support was provided by Susannah Williams. Data checks and editorial support were performed by Spirit Medical Communications Group Limited.

Funding

This SLR was sponsored by Novo Nordisk A/S.

Author information

Authors and Affiliations

  1. Wickenstones Ltd., Oxford, UK

    Adam Ben Taieb, Erika Roberts, Paulo Gomes de Freitas & Dingeman Wolfert

  2. Novo Nordisk Canada Inc., Mississauga, ON, Canada

    Maria Luckevich

  3. Novo Nordisk A/S, Søborg, Denmark

    Sara Larsen

  4. Diabetes Complications Research Centre, Conway Institute, University College, Dublin, Ireland

    Carel W. le Roux

Authors
  1. Adam Ben Taieb
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Erika Roberts
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Maria Luckevich
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Sara Larsen
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Carel W. le Roux
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Paulo Gomes de Freitas
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Dingeman Wolfert
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

ABT conducted the SLR including screening, quality control and write up of the report. ER was the second screener and undertook further quality control. ML reviewed the manuscript from Novo Nordisk. SL reviewed the manuscript and oversaw the SLR development from Novo Nordisk. As a key opinion leader CWR reviewed the manuscript and provided expert input. PGF was involved in the design and direction. DW oversaw the objectives setting, design and execution of the SLR, as well as provided oversight of the manuscript write up. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Sara Larsen.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

ML is employed by Novo Nordisk Canada Inc., Ontario, Canada, and SL is employed by Novo Nordisk A/S, Søborg, Denmark. ABT, ER, PGF, and DW are employed by Wickenstones Ltd. and received support from Novo Nordisk to complete this SLR. CWR is employed by Diabetes Complications Research Centre, Conway Institute, University College, Dublin, Ireland and did not receive support from Novo Nordisk.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Table S1.

Inclusion and exclusion criteria used in the systematic literature review. Table S2. Search strings used for the Embase search. Table S3. Search strings used for the PubMed search. Table S4. Data extraction categories. Table S5. The adapted Newcastle–Ottawa Scale. Table S6. Studies excluded following the full-text screening. Table S7. Included studies. Table S8. Quality assessment of the included studies using the adapted Newcastle Ottawa Scale.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Taieb, A.B., Roberts, E., Luckevich, M. et al. Understanding the risk of developing weight-related complications associated with different body mass index categories: a systematic review. Diabetol Metab Syndr 14, 186 (2022). https://doi.org/10.1186/s13098-022-00952-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13098-022-00952-4

Keywords

Diabetology & Metabolic Syndrome

ISSN: 1758-5996

Contact us