Bariatric surgery

Published on 16/04/2015 by admin

Filed under Surgery

Last modified 22/04/2025

Print this page

This article have been viewed 2729 times

Bariatric surgery

Richard Welbourn

Introduction

Since its inception half a century ago, bariatric surgery (from the Greek word ‘baros’ for weight or pressure and ‘-iatric’ for the medicine or surgery thereof) has grown from the preserve of a few enthusiasts to the most rapidly increasing area of surgery in developed countries. It is established that bariatric surgery is the only treatment that can result in long-lasting weight loss and improvement in obesity-related comorbidity.

Only a few operations have stood the test of time. The original jejuno-ileal bypass was abandoned due to the blind loop syndrome and resulting liver and bone metabolism problems. The vertical banded gastroplasty (VBG) has been superseded by gastric banding due to the advantage of being able to adjust the size of the gastric stoma. The gastric bypass has evolved from a continuously stapled horizontal gastroplasty and loop gastroenterostomy into a smaller pouch with a Roux-en-Y reconstruction.1,² The Magenstrasse and Mill procedure has evolved into the sleeve gastrectomy, an operation that is rapidly being taken up worldwide due to its perceived ease and lesser risk compared to gastric bypass.³ Other procedures specifically for type 2 diabetes, such as duodeno-jejunal bypass and a temporary implanted endoscopic sheath that prevents absorption in the duodenum and proximal jejunum (‘Endobarrier’), are being evaluated. The revolution of laparoscopic surgery has allowed most bariatric surgery to be performed laparoscopically with low mortality and morbidity. As a result, bariatric surgery is now established as a cost-effective intervention that should be increasingly provided.

Obesity as a public health problem

Obesity is a chronic, relapsing, debilitating, lifelong disease, recognised by the World Health Organisation as a global pandemic.⁴ The influential Foresight report estimated that by 2010 28% of women and 33% of men in the UK would be obese, rising to 50% of women and 60% of men and, even more worrying, 25% of children as well by 2050.⁵ The definitions of obesity are shown in Table 20.1. The term ‘obesogenic environment’ was used to describe the ‘results of people responding normally to the obesogenic environments they find themselves in’, given the current trends of reduced physical activity and easily available, highly advertised, relatively cheap, energy-dense foods.⁶ An example of an obesogenic food environment is the observation that the price per calorie of healthy foods (whole grains, lean meats, low fat dairy, fruit and vegetables) increased to eight times the equivalent price of unhealthy foods (sweets, calorific drinks and fatty foods) between 2004 and 2008 in Seattle, USA.⁷ Even payment methods have been implicated in unhealthy food choices: in a study in Buffalo, USA, buying of unhealthy compared to healthy foods significantly increased when the payment method was by card compared to cash (P < 0.01).⁸

Table 20.1

Definition of obesity according to body mass index

BMI (kg/m²)	WHO classification	Common clinical description
18.5–24.9	Normal range	Desirable
25–29.9	Pre-obese	Overweight
30–34.9	Obese class I	Obese
35–39.9	Obese class II	Clinically severe and complex obesity
40–49.9	Obese class III
50 and over		Super-obesity

Adapted from Dixon JB, Zimmet P, Alberti KG et al. Bariatric surgery: an IDF statement for obese Type 2 diabetes. Diabet Med 2011; 28: 628-642.

An estimated 1.46 billion adults (confidence interval (CI) 1.41–1.51 billion) globally were overweight in 2008 (body mass index (BMI) 25 kg/m² or over) and another 502 million obese (BMI > 30), including 205 million men (CI 193–217) and 297 million women (CI 280–315 million), with another 170 million children overweight or obese.⁹ In the UK, about 2% of the population, or 1.2 million people, have a BMI > 40.¹⁰

The association between obesity and the metabolic syndrome of type 2 diabetes, hypertension, sleep apnoea and polycystic ovarian syndrome is well recognised.¹¹ These, together with other harmful obesity-related comorbid disease such as non- alcoholic steatohepatosis, asthma, back and lower limb degenerative problems, cancer and depression, present a massive financial burden on health services.

The direct costs of treating these are estimated to be £5 billion per year in the UK National Health Service. This is set to double in real terms by 2050, with the indirect costs to society increasing to £50 billion.⁵

In addition, obese people are known to have higher unemployment and higher rates of claiming state benefit, and consume a disproportionate amount of the healthcare budget.^12,¹³

The traditional advice ‘eat less and exercise more’ given to patients, based on understanding of the basic energy equation, does not work in the majority: a large study in the USA of 107 000 individuals using the Behavioural Risk Factor Surveillance System found that 63% of men and 78% of women were attempting to lose weight or maintain weight at any one time. However, 34% of men and 40% of women maintained the same energy intake despite reducing fats, and only 1 in 5 of those trying to control their weight were able to eat fewer calories and do 150 minutes exercise per week.¹⁴

In chronic obesity it is recognised that patients have usually missed the boat of prevention.¹⁵ Most patients are able to diet to some extent but nearly always reach a plateau; thus a typical pattern in someone who has already become obese is a lifetime of repeated dieting and weight regain.

Dieting decreases resting energy expenditure and basal metabolic rate, powerful physiological mechanisms that counter weight loss and encourage hunger.¹⁶ In contrast, surgery enables weight loss usually to a far greater extent than can be achieved with dieting, and it allows weight loss to be maintained long term.¹⁷

It is thus not surprising that the worldwide rate of bariatric surgery has increased dramatically. In the USA, the most obese nation, the volume of surgery was 136 000 operations in 2004, while the American Society for Bariatric and Metabolic Surgery (ASMBS) estimated that 220 000 procedures were carried out in 2008.¹⁸ This figure is similar to cholecystectomy, and at this rate of surgery the UK would be doing more than 50 000 operations per year.

Worldwide the commonest bariatric procedure is the Roux-en-Y gastric bypass (RYGB). Data from the UK and Ireland National Bariatric Surgery Registry (NBSR) indicated that in 6483 operations in 2009–10, 67% of the NHS patients had a gastric bypass, 21% gastric banding and 10.5% a sleeve gastrectomy.¹⁹ In Europe the rate of gastric banding fell during the 2000s, having far exceeded gastric bypass during the 1990s.²⁰ By contrast, the popularity for banding increased in the USA after the FDA granted approval in 2001, and currently is estimated to exceed bypass (46% vs. 44%), with sleeve gastrectomy at 7.8%.²¹

Diabetes risk with obesity

Type 2 diabetes is the most important comorbid disease of obesity, affecting around 280 million people, a figure that is likely to rise to 438 million by 2030.²² About 85% of type 2 diabetics are obese and the increased risk of becoming diabetic is estimated to be 93-fold for women and 42-fold for men who are severely obese.²³ Medical treatment cannot stop diabetes from progressing and thus the epidemic is one of the largest threats to global health in the 21st century.

Cancer risk with obesity

Obesity is the second most common cause of preventable cancer after smoking. In data from the Cancer Prevention Study II in the USA from 1982 to 1998, the relative risk (RR) of death from cancer increased above a BMI of 30 for both men and women.²⁴ In 900 000 people the risk of most, if not all, cancers was increased in non-smokers.²⁵ For women with BMI > 40, there was an RR of 6.25 for uterine cancer, and the overall RR was 2.51 for other cancers, in particular kidney, cervix and pancreas. For men with BMI > 35, there was an RR of 4.52 for liver cancer, and for BMI > 30 the RR was 1.68 for all other cancers. Thus the impact of obesity appears to be important for most, if not all, organs.

Psychosocial morbidity and prejudice

Up to 60% of bariatric patients have psychiatric disorders and substantial psychiatric comorbidity.²⁶ They also have impaired self-esteem, lower quality of life, and more depression and anxiety than the general population. About one-third are victims of sexual child abuse. Anti-obesity stereotypes are common in society and also in healthcare professionals: the obese are regarded as ‘lazy, unmotivated and non-compliant’, they ‘lack self-discipline’, and they face discrimination and prejudice.²⁶ Obesity is regarded as being ‘self-inflicted’ and is associated with ‘moral failure and guilt’, underlining the need for sensitivity and non-judgmental approaches to management.

Baseline obesity-related disease

Data from 2559 patients in the US NIH-funded Longitudinal Assessment of Bariatric Surgery (LABS) study showed that obesity-related comorbid disease increases dramatically and highly significantly with rising BMI.²⁷ The P value for hypertension was 0.0018 and for diabetes, congestive heart failure, asthma, sleep apnoea, functional impairment, pulmonary hypertension, deep vein thrombosis (DVT)/pulmonary embolism (PE) risk and venous ulceration was P < 0.001 after adjustment for age, sex, race and ethnicity.

Data from the NBSR

Sixteen per cent of patients with a BMI < 40 had four or more comorbidities, and this rose to 37% for BMI 40–49.9, 47% for BMI 50–59.9 and 54% for BMI > 60 (P < 0.01). At baseline, 27.5% of patients had type 2 diabetes, 35.2% had hypertension, 18.2% had dyslipidaemia, 16.5% had obstructive sleep apnoea and 69% had impaired functional status on the basis of an inability to climb three flights of stairs without resting. Further, 53.9% of all patients reported some form of limiting arthritic symptoms. The high degree of disability in this population means that the dictum ‘take more exercise’ is not likely to be effective in weight reduction in the typical bariatric patient.

In common with the international literature, a very high proportion (80.1%) of patients undergoing bariatric surgery were female. It is not known why so few men have surgery, but men were older, heavier and had more comorbidity. For example, 43% of the men were diabetic, 50% had hypertension, 28% had dyslipidaemia and 37% had sleep apnoea (P < 0.001). The average BMI was 50.6 compared to 47.7 in females. The differences are not simply explained by the older age of males, since on age group comparison males age 40–49 had a median of three comorbidities compared to a median of two in females of the same age group (P < 0.001).¹⁹

Multidisciplinary work-up

There is strong agreement that patients should be assessed carefully by a multidisciplinary team process that includes education sessions about the different operations and aftercare (Grade A recommendation).^28,²⁹

The team should include bariatric physicians, dieticians, nurse specialists, psychologists, anaesthetists and surgeons, and support groups.29–31 Current National Institute for Clinical Excellence (NICE) guidelines, which were based on National Institutes of Health guidelines (1991), are shown in Box 20.1. It is not known what proportion of patients who fulfil the weight threshold for surgery would ever be suitable as patients must show that they are engaged and committed to the process and this is difficult to measure. Some believe that demonstration of weight loss before surgery is necessary to gauge this, but the evidence base for this is poor or lacking.²⁸ Rare endocrine causes should be excluded. Current drug or alcohol misuse and uncontrolled psychiatric disease are regarded as contraindications, as are a ‘lack of comprehension of risks and lifestyle changes required’, which might exclude patients with learning difficulties.^20,²⁸ Patients should not be operated on when pregnant.²⁸ Crohn’s disease is a relative contraindication and portal hypertension is an absolute contraindication.²⁰

Box 20.1 Criteria for bariatric surgery * (NICE Clinical Guideline 43, 2006)²⁹

• Bariatric surgery is recommended as a treatment option for adults with obesity if all of the following criteria are fulfilled:

• BMI of 40 kg/m² or greater, or between 35 and 40 kg/m² and weight-loss-responsive disease;

• all appropriate measures have failed to achieve or maintain adequate, clinically beneficial weight loss for at least 6 months;

• received or will receive intensive management in a specialist obesity service;

• fit for anaesthesia and surgery;

• commits to long-term follow-up.

• Bariatric surgery is also recommended as a first-line option for adults with a BMI > 50 kg/m².

^*SIGN (Scottish) guidelines 2010 indicate threshold for surgery is BMI > 35 and one or more weight-loss-responsive comorbidities.²³

Optimisation of patients

Multidisciplinary teams should ensure that medical comorbidities can be identified and treated, so that high-risk patients with multiple medical comorbidities can be made as fit as possible before surgery.^28,³⁰ This includes indentifying and treating obstructive sleep apnoea before surgery, and careful perioperative control of diabetes.²³ It is recommended that patients stop smoking at least 8 weeks before surgery.²⁸ Medications for epilepsy should be considered carefully as the absorption may be insufficient after gastric bypass or duodenal switch.

Obesity Surgery–Mortality Risk score (OS-MRS)

This is now established as a validated tool for gastric bypass.³² One point is scored for each of: male sex, age ≥ 45, BMI ≥ 50, presence of hypertension, known risk of DVT/PE. The possible scores range from 0 to 5. Class A (0–1 points) is low risk, class B (2–3 points) is medium risk and class C (4–5 points) is high risk. There was a 12-fold increased risk of mortality for class C patients compared to class A in the paper, which validated the score.

Thus a male patient with high BMI and hypertension, both of which are associated with central obesity, is likely to be more difficult to operate on due to the extra torque on laparoscopic instruments. Usual practice, in addition to medical work-up, is to put patients on a ‘liver shrinkage diet’ for at least 2 weeks before surgery, as this has been shown to reduce liver size, and probably reduces torque in central obesity (Grade B recommendation).^28,³³ Weight reduction before surgery may also downgrade the risk group.

Comparisons of postoperative complications and mortality between gastric banding and bypass (or sleeve) should always take the OS-MRS into account to avoid the problem of bias, i.e. comparing apples and oranges. In the NBSR, far more bypass patients were class C than banding patients.

Current bariatric operations and surgical techniques

From a surgeon’s perspective the ideal bariatric operation would provide 100% excess weight loss (EWL), complete reversal of all comorbidities, normalise functional capacity and produce no nutritional sequelae. It would also be cost-effective, provide normal quality of life and leave no loose skin. It should also need minimum resource for follow-up.

Unfortunately none provides this. A rapidly expanding literature in the last decade reflects quite variable current practice and the choice of procedure between available operations appears to be driven by patient preference, surgeon bias and local expertise.

Gastric bypass

There are several laparoscopic techniques described and there is as yet no standardisation. However, the weight loss results appear very much the same. Most agree that a short vertical lesser curvature-based gastric pouch of no more than 5–6 cm, that is separated from fundus, should be constructed, as this has been shown by MacLean and colleagues to produce weight loss over 15 years.² Starting just below the oblique fat pad on the lesser curve and usually taking the second branch of the left gastric artery, the pouch is made with linear staplers directed transversely over 2–3 cm then vertically up to just to the left of the angle of His, staying to the right of the posterior gastric artery if identified. Routing of the Roux limb can be retro- or antecolic and particular attention must be paid to the correct identification of the bowel limb to prevent ‘Roux-en-O’^34,³⁵ (Fig. 20.1).

Figure 20.1 Gastric bypass showing short vertical lesser curve-based gastric pouch with Roux-en-Y jejuno-jejunostomy.

In open surgery exact calibration of the diameter of the anastomosis of the Roux limb to the gastric pouch was thought to be critical to the rate of emptying and thus weight loss after the operation. However, as for banded gastric bypass (Fobi/Capella), there were no good data to support this.^36,³⁷ In fact, an early study by Naslund in Sweden of 29 patients found that between-patient variation in the size of the gastric outlet did not correlate with weight loss 1 year after bypass.³⁸ In the UK NBSR, the techniques of pouch-enterostomy are reported as 39% linear stapler with suture closure of the defect (after Lonroth), 23% circular stapler (after Wittgrove) and 38% hand sewn(after Higa).^19,^34,^35,³⁹

The lengths of the biliary and Roux limbs are not standardised and are impossible to measure accurately but in the US literature the biliary limb length is not usually mentioned since the assumption is that it is ‘kept short’ in order to reduce malabsorption of vitamins and minerals. The commonest description of the Roux limb is around 100–150 cm, after Brolin et al. in New Jersey and MacLean and colleagues in Montreal.^2,⁴⁰ Limited evidence suggests that a longer Roux limb of 150 cm gives more weight loss for BMI > 50, but this is not maintained in the long term. In standard bypass protein/calorie malabsorption is not the mechanism of weight loss. Since the common channel length is the determinant of available gut for absorption, and this is never measured in standard bypass, it is unlikely that longer lengths will ever be studied because of the worse nutritional consequences that would ensue.⁴¹

Gastric banding

The early perigastric approach for tunnelling the band posteriorly has now been abandoned as it is associated with high rates of band erosion into the stomach. The pars flaccida technique (through the window of the lesser omentum) is now preferred, keeping above the lesser omental bursa posteriorly. A band placed around the gastro-oesophageal junction would produce weight loss by dysphagia, and correct placement is just below, producing a small ‘virtual pouch’ of gastric mucosa above the band ²⁰ (Fig. 20.2).

Most choose to suture the band into place anteriorly with gastro-gastric tunnelling sutures in the belief that it reduces the risk of band slippage. In the NBSR, 99% of bands were placed by the pars flaccida technique (through the window in the lesser omentum) and 95% of bands were sutured anteriorly.¹⁹

Figure 20.2 Gastric band showing small ‘virtual’ pouch of stomach below gastro-oesophageal junction (a), gastro-gastric tunnellating sutures (b) and the port used for adjustment (c).

The access port is usually sutured to the rectus sheath in the upper abdomen for ease of access by a non-coring, Huber needle for band adjustments. Placement of the band is just the start of the process of weight loss. Adjustments of the band are made by injecting saline progressively in order to reach the so-called ‘sweet spot’ of optimal restriction. A patient should probably be offered monthly clinic visits, at least for the first year. One study found that patients who attended seven or more clinics achieved 50% EWL in the first year compared to only 42% EWL in those who attended six times or fewer (P < 0.01).⁴² The ability to provide good follow-up should be part and parcel of the process, but is often lacking.²⁰

Sleeve gastrectomy

Technically simpler than gastric bypass, the operation evolved from the Magenstrasse and Mill operation described by Johnston in Leeds, where the divided fundus (the ‘mill’) was left in continuity with the lesser curve-based tube (the ‘main street’).³ In the sleeve the lesser curve-based tube is constructed over a size 32 or 34Fr bougie. The dissection uses linear stapling devices and starts 3–6 cm proximal to the pylorus upwards to just lateral to the angle of His.⁴³ Some descriptions of the sleeve now remove more of the antrum, leaving a true sleeve upwards from the pylorus (Fig. 20.3).

Figure 20.3 Sleeve gastrectomy.

Sleeve gastrectomy is seen as a less risky alternative to gastric bypass, and a number of trials have now been reported (see below). The ASMBS issued an update statement on sleeve gastrectomy in October 2011, summarising the current data.⁴⁴ Himpens et al. from Belgium have the largest published series, with 5-year follow-up in which they found that 30 sleeve patients had 77.5% EWL at 3 years and 53.3% EWL at 6 years.⁴⁵ However, the starting BMI was only 39.9 and an additional 11 patients had a further bariatric procedure during follow-up, including ‘re-sleeve’ because of weight regain. Although there is much enthusiasm for sleeve gastrectomy, the lack of long-term studies and demonstrable superiority compared to both gastric bypass and banding, together with the concern about weight regain due to expansion of the sleeve, mean this operation is unlikely to replace either in the foreseeable future. The longest follow-up so far reported is 8–9 years, when 13 patients with an initial BMI of 45.8 had 68% EWL.⁴⁶

Biliopancreatic diversion/duodenal switch

Although the biliopancreatic diversion (BPD) described by Scopinaro in Naples produces greater weight loss than gastric bypass or banding, it requires very careful nutritional follow-up due to malabsorption, which is the mechanism of action of the operation.³¹ BPD appears to be done rarely; however, its duodenal switch (DS) variant, combined with sleeve gastrectomy, is performed in a few centres.⁴⁷ DS is increasingly seen as a rescue operation for weight regain after sleeve gastrectomy.⁴⁵ All patients after DS need a high-protein diet and regular vitamin and mineral supplements with monitoring for life.

The DS variant of the BPD involves a sleeve gastrectomy followed by division of the duodenum just distal to the pylorus. The ileum is divided with a linear stapler and a duodeno-ileostomy and ileo-ileostomy are made such that the common channel for food absorption measures 75–125 cm and the alimentary channel measures 100–250 cm. The long remaining biliary limb is not measured ⁴⁷ (Fig. 20.4).

Figure 20.4 Sleeve gastrectomy/biliopancreatic diversion with duodenal switch (‘duodenal switch’).

Mechanisms of bypass and banding

Both bypass and banding appear to control appetite and satiety by physiological mechanisms. In a blinded crossover trial of filling/de-filling of gastric bands, Dixon et al. demonstrated that satiety before and after eating was greater when the band was filled than when not.⁴⁸ It is proposed that stimulation of vagal afferent fibres adjacent to the band mediates the feeling of satiety, as there is no marked change, if any, in gut hormone profiles.²⁰

In contrast to banding, there has been an explosion of publications demonstrating rises in gut hormone levels after gastric bypass that favour reduced appetite and early satiety. Hormones such as peptide YY (PYY), glucagon-like peptide-1 (GLP-1) and oxyntomodulin have attracted particular attention and there are currently publications exploring changes in energy expenditure and bile acids as mediators of improved insulin resistance.^49,⁵⁰ In the foregut theory, diabetes has been proposed to be a disease of the duodenum and proximal jejunum, referring to the exclusion of this part of the anatomy to the passage of food in gastric bypass. In the original research, diabetic rats were made non-diabetic after duodeno-jejunal bypass, in which only the duodenum and proximal jejunum were excluded from the passage of food, leaving the stomach intact. This is the basis for the development of the duodeno-jejunal bypass in humans and the endoscopic duodenal sleeve technique (Endobarrier).⁵¹ If current trials show success, the endoscopic sleeve could become a useful adjunct in making patients fitter for surgery.

In the distal gut theory, rapid delivery of food from the gastric pouch via the Roux limb to the terminal ileum is proposed as the mechanism by which gut hormones rise and produce early reduction in appetite and early satiety.⁴⁹ In this study, patients with good weight loss produced high levels of PYY and GLP-1 compared to patients who did not lose as much weight (P < 0.05). In addition, in a randomised controlled trial (RCT) those receiving somatostatin after gastric bypass had attenuated gut hormone responses and experienced return of appetite compared to those receiving placebo (P < 0.05).⁴⁹ The distal gut theory contradicts the thinking behind the banded bypass.

Weight loss outcomes

A measurement of weight loss is the most consistent variable used in reporting results of bariatric surgery, being regarded by surgeons as the paramount clinical outcome. There is, however, almost complete lack of consistency about how this is measured or presented, as illustrated in Table 20.2, in which the known RCTs of weight loss outcomes from currently performed procedures are summarised.⁵² This lack of consistency leads to difficulties in combining data from studies in systematic reviews and meta-analyses, and it makes comparisons between studies and centres difficult. Most of the available evidence consists of case series (level III evidence).

Table 20.2

Summary of RCTs of currently performed procedures ⁵²

Adv/Ev, adverse events; AGB, adjustable gastric banding; BPD, biliopancreatic diversion; DS, duodenal switch; EWL, excess weight loss; Mini-RYGB, Mini-gastric bypass; NR, not recorded; NS, not significant; RYGB, gastric bypass; SG, sleeve gastrectomy; TBWL, total body weight loss. N.B. Karamanakos and Kehagias report data from the same RCT.

^*Added in addition to Padwal systematic review.

^**P < 0.05.

Although the long-term outcome of patients who have had gastric bands is not known, one systematic review suggests that if the band is correctly placed, the patient is well followed up and does not suffer a complication, the weight loss at 3–4 years may be similar to gastric bypass, but without the risk of the initial operation.⁶¹ In both RCTs (see below) weight loss was better for bypass.

Enthusiasts of gastric banding maintain that since these patients lose their weight slowly over 3–4 years after surgery, differences in early EWL compared to other operations are not relevant to long-term health (although this has not been studied). The reality is that follow-up outside of funded studies is usually extremely poor, despite sometimes strenuous attempts by clinicians to make contact. In the O’Brien study, data from 3874 gastric bypass patients and 10 041 gastric band patients were analysed at 5 years, and at this time the weight status of only 4.5% of bypasses and 6.4% of bands was known. Even so, the consensus is that this is a good estimation of weight loss outcomes in practice, with some weight regain after the nadir of weight loss being usual after bypass. Most patients settle at around 55–65% EWL long term, perhaps 10% more EWL than the average band patient.

A challenge of managing gastric band patients is how to help the proportion of patients who fail to lose much weight. In the two RCTs discussed below, failure as defined by BMI > 35 was found in the Italian trial in 4.2% of bypasses and 34.6% of bands (P < 0.001). In the Californian RCT patients failing to lose the first quintile of excess weight were none for bypass and 16.7% for bands (P < 0.05). Many patients in this position opt for revision to a gastric bypass if possible. An equally challenging group is the bypass patients who have significant weight regain at 3–4 years after the nadir of weight loss. Treatment options include adding a gastric band to the existing gastric pouch.

Band versus bypass RCTs

Two RCTs have compared gastric bypass and adjustable gastric banding. Angrisani et al. in Naples randomised 24 patients to gastric bypass and 27 patients to banding.⁶² There was no mortality and one band patient was lost to follow-up. Four band patients required revision to other bariatric operations during follow-up, two for gastric pouch dilatation and two for unsatisfactory weight loss. Three bypass patients required early re-operations because of life-threatening complications (12.5%). Fifteen bypass patients (62.5%) achieved BMI < 30 compared to three (11.5%) band patients (P < 0.001). The authors concluded that gastric bypass produces better weight loss and fewer failures, acknowledging that the high complication rate was accounted for by their learning curve in the bypass operation.

Nguyen et al. in California randomised 111 patients to gastric bypass and 86 to banding.⁶³ There was no mortality in either group. At 4 years patients were grouped into quintiles of percentage excess weight loss (% EWL). The highest quintiles representing 60% EWL or more had 64.1% of bypass patients and 15.3% of band patients. The lowest quintiles representing < 40% EWL had 5.1% of bypass patients and 50.0% of band patients (P < 0.05). Although the authors concluded that weight loss at 4 years was superior after bypass, the follow-up rates achieved were only 83.1% for bypass and 93.3% for band, and no explanation was given for the different numbers randomised to each arm.

Given that these two procedures are the commonest in the NHS (and worldwide), the methodological weaknesses of the two RCTs limit their general applicability. A new trial comparing these head to head has recently been funded. The By-Band trial, which opened in 2012, is a multicentre RCT that aims to randomise 760 patients in the UK. The trial is designed to answer the question of whether gastric bypass is better than gastric banding in terms of health-related quality of life and at least as good as gastric banding in terms of achieving weight loss. Both aspects of this dual primary end-point need to be achieved to be certain that bypass is superior to gastric banding. They will be measured at 3 years. A cost-effectiveness analysis will also be included and it is hoped that the results will inform the debate on how bariatric services should be provided.

Band versus sleeve gastrectomy RCT

In the largest RCT, Himpens et al. in Belgium randomised 80 patients to either operation.⁵⁴ Weight loss was better for sleeve at each time point measured: 57.7% EWL vs. 41.4% EWL at 1 year and 66% EWL vs. 48% at 3 years (both P < 0.01). Although there were several operations for complications in each group, an additional two patients in each was converted to another bariatric operation due to lack of weight loss.

Banded versus non-banded gastric bypass RCT

There has been one RCT of banded versus non-banded (standard) gastric bypass. The banded gastric bypass involves placing a silicone ring around the gastric pouch above the pouch-enterostomy anastomosis, with the intention of causing a fixed amount of restriction and thus greater weight loss. In an RCT, Bessler et al. randomised 46 patients to banded bypass and 44 to standard bypass.⁵⁶ There were no significant differences in EWL at 6, 12 and 24 months (43.1% vs. 24.7%, 64.0% vs. 57.4%, and 64.2% vs. 57.2%). However, there was a small advantage at 36 months for the banded group (73.4% vs. 57.7%, P < 0.05), although the number available for follow-up was not specified. Other than in a few centres, this operation has not received widespread uptake and the proposed mechanism of action (hold-up of food in the pouch) is at variance with the distal gut theory.⁴⁹

There has been one RCT of standard gastric bypass versus the so-called mini-gastric bypass. In this variant a long gastric pouch/tube (midway in length between the pouch in standard gastric bypass and that in a sleeve gastrectomy) is created and an antecolic loop gastroenterostomy is made without a Roux-en-Y. Although the weight loss outcomes are reported to be similar to standard gastric bypass, it is an uncommon operation and there has been reluctance to recognise it for credentialing purposes by the ASMBS and the American College of Surgeons (ACS).^31,⁵³

Sleeve versus bypass RCTs

There are at least three RCTs of sleeve gastrectomy versus gastric bypass and one of sleeve versus mini-gastric bypass. Kehagias et al. in Greece have published the longest follow-up in an RCT of bypass versus sleeve and they found no significant difference in weight loss between the two at any time point up to 3 years (3-year data: 68% EWL for sleeve and 63% EWL for bypass).⁴³ An earlier report of the same trial found that weight loss at 1 year was similar for the two operations and this is the general consensus for medium-term comparison of bypass with sleeve.⁵²

Bypass versus duodenal switch RCT

There is one published trial, from Norway/Sweden.⁵⁹ In this study, Søvik et al. found much greater weight loss after sleeve gastrectomy with DS compared to bypass at 2 years. Cholesterol and other measures of lipid metabolism were improved more by duodenal switch but measures of vitamin D metabolism were worse. Although quality of life was broadly similar in follow-up between the two, adverse events were more frequent after DS (62% vs. 32%, P < 0.02). These were mainly due to the unavoidable nutritional consequences of DS and this illustrates the main challenge in follow-up after this operation.

Complications of surgery

Operative mortality

Published mortality rates after gastric banding suggest it is very safe, with about a 0.05–0.1% risk of dying. The published mortality from gastric bypass is higher at about 0.2%, with the mortality from sleeve gastrectomy between the two. The data in the table refer to large published national databases or registries reporting since 2009 (Table 20.3).^19,^21,^64,⁶⁵

Table 20.3

Operative mortality data from UK&I National Bariatric Surgery Registry, Hospital Episode Statistics data, Longitudinal Assessment of Bariatric Surgery (USA) and American College of Surgeons Bariatric Surgical Center Network ^19,^21,^64,⁶⁵

HES, LABS and ACS data were P < 0.05 between groups. In the LABS study data on 117 sleeve gastrectomy patients were not analysed further. N/A, not available.

^*Total laparoscopic/open.

^†Laparoscopic only.

Open bypass has higher mortality (data not shown), and the Hospital Episode Statistics (HES) data from the UK were from 2000 to 2008, when much of the surgery was not yet laparoscopic. By far the largest report of outcomes is from the Bariatric Outcomes Longitudinal Database of the ASMBS Centers of Excellence programme, in which 57 918 patients had gastric bypass (54.7%), banding (39.6%) or sleeve gastrectomy (2.3%).⁶⁶ The 90-day mortality rate for those with follow-up was 0.22%, but neither mortality per procedure nor 30-day re-operation rates were presented. The ACS Bariatric Surgery Center Network reported higher morbidity, re-operation and re-intervention rates for gastric bypass than for banding, with sleeve between the two.²¹ Similarly, the Michigan Bariatric Surgery Collaborative reported 30-day complication rates for bypass of 3.6% compared to 0.9% for gastric banding and 2.2% for sleeve.⁶⁷

Gastric bypass

Early

Anastomotic leak, particularly at the pouch-enterostomy, is a feared and potentially deadly complication and may be present in up to 3%.⁶⁸ It typically occurs within 24 hours, i.e. the anastomosis is unlikely to have been water-tight initially. It is essential that a patient who has abdominal pain and a sense of ‘impending doom’ be considered for same-day re-laparoscopy or contrast-enhanced computed tomography (CT) scan.⁶⁸ Tachycardia is an unreliable sign, particularly when the patient is taking beta-blockers for hypertension. The pouch-enterostomy is more likely to leak than the jejuno-jejunostomy and, because of its accessibility, it is usual to do a leak test during surgery. Same-day re-operation and direct suturing or T-tube placement and drainage are essential due to the limited reserve in patients with severe comorbidity. Poor healing is characteristic and patients are severely catabolic, with very high energy requirements.

Bleeding from anastomoses and staple lines may present postoperatively as melaena in up to 4% and almost always settles, sometimes needing transfusion.⁶⁹ Closed loop obstruction, particularly if there is a ‘Roux-en-O’ or acute internal hernia, is life threatening and must be excluded or rectified quickly.

DVT/PE is still the commonest cause of mortality after bariatric surgery, probably accounting for half the deaths. It should be routine to give appropriate prophylaxis with calf compression stockings and inflatable leggings during surgery and at least one postoperative week of appropriate anticoagulant for any operation other than banding.⁷⁰ All patients should be encouraged to mobilise on the same day of surgery and enhanced recovery regimens are common. Data from the UK NBSR showed that > 90% of banding patients are discharged home by 24 hours and > 80% of bypass and sleeve patients are discharged home by postoperative day 3.¹⁹

Late

Stricture at the pouch-enterostomy occurs in up to 5%, probably depending on the technique of anastomosis and any tension.³⁵ It usually responds to endoscopic dilatation. Marginal ulcer is reported to occur and it may be less common if Helicobacter pylori infection is sought and treated preoperatively. Internal herniation with bowel ischaemia is a known serious risk as the hernia spaces made by the anatomical rearrangement enlarge as weight is lost. It is mandatory for every patient to be on lifelong vitamin and trace mineral supplementation due to the unavailability of the duodeno-proximal jejunal segment for absorption (see below).

Most experts recommend prophylactic suturing of the Petersen defect (space behind the Roux limb), the mesocolic defect (in retrocolic routing) and the jejuno-jejunostomy defects with non-absorbable sutures, although there are no data to show this makes a difference.³⁵ An RCT based on the Swedish Registry (SOREG) is currently ongoing.

The risk of symptomatic internal hernia is at least 2% over 1–2 years after surgery.⁷¹

Gastric Band

Early

The re-operation rate within 30 days after banding is about 1%. It is considered mandatory to give prophylactic antibiotics. Early infection of the gastric band port wound is a potential disaster as it may indicate that the whole band is infected. After a suitable course of antibiotics a pragmatic approach (not evidence based) is to re-laparoscope and cut the band tubing as it exits the abdominal wall, completing this part of the operation before opening the infected port wound and removing the port. If the infection is localised to the port wound only, it may be possible to reconnect another port at a later date.

A complication that can occur at any time after the band has been filled is acute obstruction due to a food bolus or over-filling. Every hospital should have an ‘emergency de-fill box’ for out-of-hours use containing a non-coring Huber needle for this purpose.⁷² Patients should be educated about this possibility and know how to contact their bariatric team.

Late

Every patient must be carefully counselled about the long-term risk of re-operation. Suter et al. in Switzerland found there was a 21% cumulative re-operation rate over 7 years.⁷³ The real-life re-operation rate is not known since so many patients are lost to follow-up, but a realistic estimation may be 10–20% over 5–10 years. Re-operations may be minor, e.g. a needle stick injury to the tubing adjacent to the port or failure of the port membrane causing leakage and requiring local attention to replace the port or shorten the tubing. Much more serious complications are infection, requiring removal of the whole band, slippage of the position of the band around the upper part of the stomach, and erosion, where the band migrates into the lumen of the stomach.

Slippage: This typically presents at 2–3 years, and produces vomiting and loss of restriction. This is usually a non-urgent situation and treatment consists of de-filling the band, a contrast X-ray to make the diagnosis and laparoscopic repositioning of the band. Concentric pouch dilatation, sometimes with a dilated oesophagus and lack of weight loss, can be confused with slippage and may indicate lack of a satiety response from the band.

Slippage can be a life-threatening emergency if there is severe abdominal pain, indicating ischaemia of the gastric pouch above the band and resulting necrosis. Treatment consists of emergency de-filling, a contrast X-ray and urgent laparoscopy. If possible, the band should be repositioned. There is no place for waiting for endoscopy to make the diagnosis.⁷²

Concentric pouch dilatation, sometimes with a dilated oesophagus and lack of weight loss, can be confused with slippage and indicate lack of an appetite/satiety response from the band.

Erosion: This occurs in about 1% of patients long term and may result from, or be the cause of, port-site infection. It does not present as an emergency. Sudden loss of restriction is the norm, and a red port scar de novo is worrying for erosion. Endoscopy or contrast X-ray provides the diagnosis and band removal is usually done laparoscopically, although it can be done by endoscopy. Revision surgery, usually to convert to a gastric bypass, should be delayed for at least 6 months to allow healing.

Sleeve gastrectomy

The main early complication of sleeve is leakage along the staple line, usually at the angle of His. This is thought to be due to the relative back pressure produced by an intact pylorus against oesophageal peristalsis. With an incidence reported consistently at up to 3%, leaks can take months to heal, and many surgeons routinely use reinforcement materials or continuous suturing along the staple line to try to prevent a leak.⁷⁴ Treatment options include percutaneous drainage, endoscopic stenting, parenteral nutrition/jejunal feeding, proton-pump inhibitors and fibrin glue.

High-volume specialisation

Data from the Michigan Bariatric Surgery Collaborative assessed hospital complication rates in 15 275 patients having bypass, banding and sleeve gastrectomy in 2006–2009. The overall risk for all procedures (after adjusting for comorbidity status and OS-MRS risk; see above) of serious complication was 4.0% (CI 2.8–5.3) for annual volumes of < 150 cases per hospital and < 100 cases per surgeon combined, compared to 1.9% (CI 1.4–2.3) for ≥ 300 cases per hospital and ≥ 250 per surgeon combined (P < 0.001).⁶⁷ In addition to specific surgical complications, the team approach characteristic of high-volume hospitals appears to facilitate better outcomes (Grade D recommendation).^30,³¹ Most experts recommend round-table discussion of patients on a regular basis. Since the success of banding is completely reliant on good follow-up, the multidisciplinary team should be structured to enable this.³¹ In addition, the International Diabetes Federation statement (see below) made the following comment:

‘Bariatric surgery should be performed in high-volume centres with multidisciplinary teams.’²³

There are a number of other case series supporting high-volume specialisation in gastric bypass surgery that also recognise the high risk of the learning curve and the need for mentoring for this operation.⁷⁵

Comorbidity outcomes

Data from a meta-analysis of 21 studies showed improvements in diabetes, hypertension, dyslipidaemia and sleep apnoea in nearly all patients.⁷⁶ Numerous other studies show improvements in individual organ failures. For example, liver biopsies on 36 patients were assessed blindly before and 2 years after gastric banding.⁷⁷ Histology showed marked improvements in indices of grade and stage of non-alcoholic steatohepatosis (P < 0.01). At follow-up only three patients had a fibrosis score of 2 or more compared with 18 patients at baseline (P < 001). Another example is cardiac function; in one study 423 gastric bypass patients were compared to 733 non-operated controls with an echocardiogram at baseline and at 2 years. Left ventricular mass and right ventricular cavity area had both decreased at follow-up (P < 0.01), and left atrial volume stayed the same in bypasses but had increased in controls (P < 0.05), all indicating improved cardiac function.⁷⁸

Swedish Obese Subjects study

The largest series of detailed long-term comorbidity data is the Swedish Obese Subjects (SOS) study (Table 20.4).⁷⁹ This study is also probably the best known in bariatric surgery. It was initiated in 1987, recruiting to 2001, and is a non-randomised cohort study of the effect of bariatric surgery versus medical therapy. At the time, before laparoscopic techniques, bariatric surgery was considered too dangerous to allow ethical randomisation between surgery and best medical therapy. Thus, 2037 patients choosing not to have surgery were matched and compared to 2010 surgical patients. Over a mean follow-up of 10.9 years (range 4.9–18.2) there was no significant weight loss in the medical group, who received all appropriate therapy for their comorbid disease. In contrast, weight loss in the surgical groups, consisting of non-adjustable gastric banding, gastric bypass and vertical banded gastroplasty, was between 13.2% and 25%. There were improvements in every variable studied, but it was noted that hypertension tended to recur, associated with advancing age of the study population.

Table 20.4

Incidence of comorbidity: 10-year data from the Swedish Obese Subjects study ⁷⁹

Reproduced from Sjöström L, Lindroos A-K, Peltonen M et al. Lifestyle, diabetes, and cardiovascular risk factors 10 years after bariatric surgery. N Engl J Med 2004; 35:2683–93. With permission from Massachusetts Medical Society.

Data from the NBSR

The proportion of patients recorded as showing no indication of comorbidities at 1 year in the NBSR is shown in Table 20.5.¹⁹ The functional improvement was dramatic and at 2 years the rate of diabetes had fallen by 85.5%. Most of the effect of surgery in the NBSR results was from gastric bypass; in fact, for banding only dyslipidaemia and functional impairment showed significant improvement at 1 year, although all other variables showed non-significant improvement. The data are also consistent with the slower rate of weight loss universally found with gastric banding compared to RYGB.²⁰ Because of a lack of well-controlled randomised studies comparing the two operations, it is not clear whether the apparently slower rate of improvement in comorbidity with banding translates to different outcomes in the medium to long term.

Table 20.5

Incidence of comorbidity: 1-year data from the UK NBSR (patients with complete follow-up, all operations)¹⁹

GORD, gastro-oesophageal reflux disease; PCOS, polycystic ovarian syndrome.

Quality of life (QoL) after bariatric surgery

There are many studies indicating that quality of life improves after surgery compared to no surgery. Again, the largest study is the SOS, which reported that all QoL measures improved in 655 surgery patients compared to 621 controls when assessed at 2–10 years.¹⁷ Variables included General Health Rating Index, Sickness Impact Profile, Mood Adjective Check List, and Hospital Anxiety and Depression score. Improved QoL was also found in the RCT by O’Brien et al. where band patients had improvement at 2 years in five of eight domains in the SF-36 tool – physical function, physical role, general health, vitality and emotional role.⁶¹ Improvement in psychological functioning is not universal, however, as it also appears that there is an increased risk of suicide after surgery.⁸⁰

Patient-reported outcomes

There is wide variation in the questionnaires used to assess health-related quality of life after bariatric surgery, with no consensus as to how the results should be reported. Poor study designs and lack of reporting integrated with clinical outcomes means there are few examples of how patient-reported outcomes have influenced practice and no data on which domains of health are important. As a result it has not been possible to produce a meta-analysis of quality-of-life studies. Thus there is a need for a core outcome set for bariatric surgery reporting that includes this variable.

Diabetes outcomes

Bariatric surgery offers very substantial benefit for diabetes. In the RCT by Dixon et al. of gastric banding versus intensive medical therapy for recent-onset diabetes, HbA1c was lowered more with banding (7.8 ± 1.2 preop vs. 6.0 ± 0.82 postop, − 1.81 ± 1.24 percentage points) than with medical therapy (7.6 ± 1.4 preop vs. 7.21 ± 1.39, − 0.38 ± 1.26 percentage points, P < 0.001). Also, there was better remission in the surgery group at 2 years (73% vs. 13%, P < 0.001) and significantly fewer people with the metabolic syndrome (70% vs. 13%, P < 0.001).⁵⁷

In the systematic review of outcomes in 4070 diabetic patients by Buchwald et al., it was estimated that remission at 2 years was achieved in 57% of patients after gastric banding, 80% after gastric bypass and 95% after biliopancreatic diversion.⁸¹ Also, HbA1c and fasting glucose levels fell significantly, indicating better diabetes control. However, the main drawback of such a meta-analysis is lack of consistent reporting of the terms used to define remission.

Given that bariatric surgery is presently only being offered to a very small proportion of those who could benefit, even in countries with highly developed healthcare, the International Diabetes Federation perceived the need in 2011 to offer international guidance.²³ The recommendations included: ‘bariatric surgery is an appropriate treatment for type 2 diabetes and BMI ≥ 35 not achieving recommended treatment targets with medical therapy, especially where there is other obesity-related comorbidity’ (author’s italics).

Various terms for improvement of diabetes control after surgery have been used, including ‘resolution’ and ‘cure’, but ‘remission’ is now the preferred term. In 2009 the American Diabetes Association introduced a more strict definition of complete remission as being a return to normal measures of glucose metabolism (HbA1c < 6%, fasting glucose < 5.6 mmol/L) in the absence of hypoglycaemic medication over at least 1 year after bariatric surgery.⁸² One study has assessed remission rates with the new definition. At 2-year follow-up in 209 diabetics, HbA1c was reduced in all surgical groups (P < 0.001), and complete remission rates were 40.6% after gastric bypass, 7% after gastric banding and 26% after sleeve gastrectomy.⁸³ Although the study was not randomised, the differences between the procedures were significant (P < 0.001).

The study by Dixon et al. in 2008 is the only randomised trial comparing diabetes outcomes between banding and intensive medical therapy. Recently, two further studies have provided randomised evidence for gastric bypass, sleeve gastrectomy and BPD as well compared to medical therapy. Both RCTs highlight the main therapeutic aim of controlling HbA1c rather than purely focusing on the ability of surgery to reduce medication usage. In the RCT by Schauer et al., HbA1c fell from 9.3 to 6.4 with gastric bypass, and 9.5 to 6.6 with sleeve gastrectomy (2.9 percentage points for both).⁸⁴ In comparison, HbA1c only fell from 8.9 to 7.5 (1.4 percentage points, both P < 0.001) with intensive medical therapy. The primary end-point (proportion with HbA1c < 6.0 at 1 year) was achieved by 42% with gastric bypass, 37% with sleeve gastrectomy and 12% with medical therapy (both P < 0.001 compared to medical therapy).In the RCT by Mingrone et al., the starting HbA1c was 8.65 ± 1.45, and this fell to 6.35 ± 1.42 with gastric bypass, 4.95 ± 0.49 with BPD and 7.69 ± 0.57 with intensive medical therapy (P < 0.01 between each group).⁸⁵ The primary end-point (fasting glucose < 5.6 mmol/L and HbA1c < 6.5% on no medication) was achieved by 75% with gastric bypass, 95% with BPD and none with medical therapy (P < 0.001 between groups). These are the only randomised data between bypass and BPD, and suggest that the latter gives better glycaemic control. There are no randomised data studying the effect of DS on HbA1c.

The different rates of improvement/remission between operations supports the consensus that bypass (and BPD) has effects other than pure weight loss on improving glycaemic control. It is widely presumed that this is due to the postoperative rise in gut hormones such as the incretin GLP-1, which is associated with rapid improvement in diabetic control in the immediate postoperative period before significant weight loss.⁸⁶

While all of the currently popular operations dramatically improve glycaemic control, many patients can still expect to be on diabetic medications postoperatively. Previously, surgeons tended to champion reduction in medication usage and to gauge the success of surgery in these terms. Focusing on glycaemic control (HbA1c) as well should not detract from the very substantial financial benefit of patients coming off treatment (see below). Thus, surgery is ‘a component of the ongoing treatment of chronic disease management of type 2 diabetes and obesity’ and patients still need long-term follow-up by interested physicians.²³

Nutritional support in follow-up

All bariatric patients should have routine metabolic and nutritional monitoring life-long (Grade A recommendation) as it is increasingly recognised that obese patients have pre-existing nutritional deficiencies that may be exacerbated by bariatric surgery, especially gastric bypass and duodenal switch.²⁸

The frequency of assessment should be at least 3- to 6-monthly in the first year if there are known deficiencies, at least 6- to 12-monthly in the second year and at least annually thereafter.²⁸

Long-term survival benefit after surgery

Several population-based studies now indicate that bariatric surgery confers survival benefit. The SOS study was the first prospective study to show that bariatric surgery confers survival benefit; even after the 90-day mortality rate from surgery of 0.25%, 129 control patients died compared to 101 in the surgical group, hazard ratio (HR) 0.76 (95% CI 0.59–0.99), with a time to reach significance (P = 0.04) of 13 years.⁸⁷

Another study that also reported in 2007 was a retrospective analysis of prospectively collected data.⁸⁰ Using self-reported BMI data collected from driving licences in Utah, Adams et al. were able to match, for age, sex and BMI, 7925 patients who had undergone gastric bypass with 7925 controls. The average age was 39.5 and BMI 45.3, but it was not possible to collect data on pre- or postoperative comorbidity or BMI at follow-up. The study period was 1984–2002 and the mean follow-up was 7.1 years. Expressing mortality as deaths/10 000 patient years, 37.6 patients died in the years after surgery compared to 57.1 controls (40% reduction, P < 0.001). Disease-specific reductions in mortality were also seen for coronary artery disease (56% reduction, 2.6 vs. 5.9, P = 0.006), diabetes (92% reduction, 0.4 vs. 3.4, P = 0.005) and cancer (60% reduction, 5.5 vs. 13.3, P = 0.001). The post-surgical mortality at 1 year was 0.53%, which compared to 0.52% of controls dying in the same period.

The reports described confirmed survival benefit after mainly gastric bypass surgery. O’Brien and Dixon’s group in Melbourne has also reported survival benefit after gastric banding.⁸⁸ In a retrospective series of 966 operated patients followed up for 4 years, the HR for death was 0.28 (95% CI 0.10–0.85) compared to a matched cohort of 2119 community controls followed up for 12 years.

The first systematic review and meta-analysis of long-term mortality outcomes after mainly gastric bypass and banding was published in 2011.⁸⁹ Eight studies were included, totalling 14 052 surgical patients and 29 970 controls, with follow-up of between 2.5 and 12 years. The odds ratio (OR) for mortality after surgery was 0.55 (CI 0.49–0.63, all eight studies), the OR for cardiovascular mortality was 0.58 (0.46–0.73, four studies) and the OR for non-cardiovascular mortality was 0.70 (0.59–0.84, four studies).

To date there are no data on survival benefit from other bariatric operations. However, a more recent study of male gastric bypass patients in Veterans Administration hospitals in the USA, who were predominantly older than in the studies above, also reported in 2011.⁹⁰ No survival benefit at 6.7 years was found in the 850 operated patients compared to 847 controls. This is the first study not to show survival benefit after bariatric surgery, possibly because of relatively high mortality in the operated patients and the relatively short follow-up.

After surgery, patients may not return to the level of risk of the general population. In Sweden 13 270 bariatric surgery patients still had an HR for mortality of 1.24 (CI 1.15–1.34) when compared to 132 700 individuals from the general population who did not have surgery.⁹¹ One of the many unknowns regarding the mechanism by which surgery reduces mortality is whether any of the operations halt or reverse the characteristic microvascular changes of diabetes.

Cancer incidence after bariatric surgery

A further finding from the SOS study is the reduction in incidence of all cancers in follow-up. Using the Swedish National Cancer Registry to observe cancer incidence from 1987 to 2005, Sjöström et al. found that 117 cancers developed in 2010 operated patients compared to 169 cancers in 2037 controls (HR 0.67, P < 0.0009).⁹² Although the absolute numbers are small, Sjöström et al. found it useful to compare this effect of surgery to statin treatment, where the HR for development of fatal or non-fatal myocardial infarction was 0.80 in > 90 000 patients.⁹³

The Utah driving licence researchers found a similar reduction in incidence of cancer diagnoses during an average of 12.5 years follow-up in 6596 gastric bypass patients when compared to 9442 severely obese controls.⁹⁴ Using the Utah Cancer Registry, Adams et al. found the HR for cancer was 0.76 (CI 0.65–0.89, P = 0.0006). The reduced incidence of cancer almost entirely comprised the female known obesity-related cancers; in particular, there was a 78% reduction in endometrial cancer (HR 0.22, CI 0.13–0.40, P < 0.0001).

The risk of dying from all cancers was also decreased by 46% (HR 0.54, CI 0.37–0.78, P = 0.001). Interestingly, the cancer mortality was also decreased in non-obesity-related cancer by 47% (HR 0.53, CI 0.31–0.91, P = 0.02).

Cost-effectiveness of bariatric surgery

Large studies from the UK, USA and Canada have recently evaluated the economics of bariatric surgery. In 2009 the HTA programme published a systematic review in which Picot et al. analysed 5386 journal articles.⁹⁵ Of these, 26 were used in the clinical effectiveness review, including 23 RCTs. Bariatric surgery was found to be more clinically effective for weight loss than non-surgical methods, with some measures of quality of life improving but not others. In addition, cost-effectiveness was estimated using incremental cost-effectiveness ratios (ICERs) per quality-adjusted life year (Table 20.6). Particularly for longer time horizons up to 20 years, these ratios were deemed cost-effective, although not for the BMI group 30 + and < 35.

Table 20.6

Data from UK Health Technology Assessment report 2009 ⁹³

Another, updated systematic review has recently reported and expressed health economic outcomes as incremental cost-utility ratios (ICURs) in 2009 US dollars.⁵² ICURs ranged from $1000 to $40 000 per quality-adjusted life year (QALY), and it was not possible to distinguish between different procedures. O’Brien’s group has analysed the lifetime projected cost savings for recently diagnosed diabetics treated with gastric banding compared to medical therapy from their RCT, and found surgery cost less and led to more QALYs.⁹⁶ Using a different model in the USA that is based on a meta-analysis, other health economists found the cost-effectiveness over a 10-year timescale for gastric banding to be between $US11 000 and 13 000/QALY for patients with diabetes.⁹⁷ Gastric bypass was found to be more cost-effective at between $US7000 and 12 000/QALY for diabetes, and all of these estimates were well below the threshold of cost/QALY usually used to justify expenditure.

In another study the 3651 patients in a US employer claims database of 5 million who were operated between 1999 and 2005 were matched to similarly obese non-operated controls with similar baseline comorbidity using the ICD9 code 278.01.⁹⁸ The operations were mainly gastric bypass. Total healthcare costs including hospital visits and medications were recorded during the 6 months before surgery and continuously up to 5 years. Using a model adjusted for inflation, Cremieux et al. estimated that all costs were recouped for laparoscopic bariatric surgery within 2 years and for open bariatric surgery within 4 years.

In a study using a US insurance database of 8.5 million people, Klein et al. analysed the pre- and postoperative drug costs in diabetic patients (n = 808) matched to non-operated controls (n = 808).⁹⁹ They found that at 3 months after surgery insulin usage had fallen to 43% in operated patients and stayed at 84% in controls. Medication and supply costs of insulin at this time were $US33 and $US123, respectively (both P < 0.001). At 6 months only 28% of operated patients still had a diagnosis of diabetes and it was found that the break-even point for the cost of surgery due to the cessation of diabetes medications alone was 26 months.

Despite the assumed benefits, the rate of bariatric surgery for the NHS in England in 2009–10 was approximately 0.30% of those who might benefit (3642 of 1.2 million). The availability of surgery appears to be no greater in other countries with a similar scale of obesity, with healthcare providers seemingly caught by the imperative to balance budgets within a 1-year timescale given that the cost of surgery may not be recouped until at least 1–2 years.¹⁰⁰ The case for diabetes is an illustration of this.⁹⁹ It is not known how much the willingness of healthcare systems to provide this surgery might be influenced by societal prejudice.

Wider economic benefit of bariatric surgery

The above cost-effectiveness models do not include the benefits to society as a whole, such as gainful employment after surgery. A UK Office of Health Economics (OHE) report used a novel approach in which they estimated the expected gains arising from unemployed patients going back to work.¹⁰¹ The estimate was based on a study of patients who at a median 14 months follow-up after surgery increased their paid hours worked by 57% and reduced their state benefit claims by 75%.¹⁰² The model found that if 25% of eligible patients (140 000 in their estimate) received surgery the boost to the GDP would total £1.295 billion at 3 years due to increases in paid employment, with an additional £151 million being returned to the economy by reducing benefits costs. Although this was a single study, reports from at least three other European countries, including the SOS study, have also shown increases in paid work after surgery.^103–105 If the additional economic factors alone are considered, without including any cost savings from medication reduction or less hospitalisation, surgery pays for itself within 1 year.¹⁰¹