Clinical and cost-effectiveness of cholecystectomy versus observation/conservative management for preventing recurrent symptoms and complications in adults presenting with uncomplicated symptomatic gallstones or cholecystitis: a systematic review and economic evaluation

Introduction

Gallstone disease (cholelithiasis) is one of the most common and costly gastrointestinal disorders in industrialised societies. 1–4 The prevalence of gallstones in these adult populations is approximately 10–15%. 1,5–9 Gallstones are more common in women and people over the age of 40 years. 10

Approximately 80% of people with cholelithiasis do not have symptoms,11,12 can remain asymptomatic for many years and do not require treatment. 13,14 About 20% of people experience pain and clinical complications. 1,5,6,15

Surgery remains the treatment of choice for symptomatic gallstones,16 even though early natural history studies showed that recurrent pain attacks may diminish in up to half of symptomatic people. 17 As many studies have concentrated on the best timing of performing surgery and on operative outcomes and complications, the question whether or not cholecystectomy is always required in people with mild, uncomplicated symptomatic gallstones has not been rigorously evaluated. There is evidence to show that it is feasible to conduct randomised controlled trials (RCTs) to compare the effects and safety of cholecystectomy with those of observation/conservative treatment. 18

The aim of the present assessment is to evaluate the clinical effectiveness and cost-effectiveness of cholecystectomy compared with observation/conservative management for preventing recurrent symptoms and complications in adults presenting initially with uncomplicated symptomatic gallstones for whom surgery is considered a treatment option.

Aetiology, pathophysiology and clinical presentation

The gallbladder is a pear-shaped pouch which stores and concentrates bile entering the organ via the hepatic and cystic ducts (Figure 1). The concentrated bile is later ejected into the duodenum in response to food entering the intestines. The gallbladder is located beneath the liver and can hold between 30 and 50 ml of bile. Histologically, the gallbladder wall consists of serous, muscular and mucous layers. 19 Gallstones are solid accumulations of either cholesterol, mucin, calcium bilirubinate or protein20 and occur when the constituents come out of solution. 21 Most gallstones (> 80%) consist largely of cholesterol,7,22 whereas pigment stones consist of calcium bilirubinate and mucin glycoproteins. Both types of gallstone form in the gallbladder itself.

FIGURE 1.

Biliary system. Source: http://www.daviddarling.info/encyclopaedia/G/gall_bladder.html (last accessed May 2013). Reproduced with permission of David Darling.

Known risk factors for adults with gallstone disease include:8,23

• increasing age

• obesity

• family history of gallstone disease

• ethnicity

• recent weight loss (for example, crash diets or weight-loss surgery)

• digestive disorders (e.g. Crohn’s disease, irritable bowel syndrome)

• cirrhosis

• diabetes.

Women who:

• take oral contraception

• have high-dose oestrogen therapy or

• are pregnant

are at higher risk (twice as likely as men) of developing gallstones. Some risk factors are modifiable, such as diet and obesity, while others are not (e.g. sex, age, concomitant diseases). 22

The most relevant symptom of gallstone disease is pain (i.e. biliary colic). Typically, biliary colic is defined as pain, moderate to severe, localised to the right upper quadrant or epigastrium lasting more than half an hour which occasionally can radiate to the right scapula. The pain may be accompanied by nausea and vomiting, may occur post-prandially and may wake the person at night. Although biliary pain is specific to gallstone disease, many people may present with other abdominal symptoms. 24 In some people, the symptoms are mild and consist of vague indigestion or dyspepsia. Dyspeptic symptoms (e.g. belching, bloating, heartburn), however, are not specific to gallstone disease. 5,7,25

Epidemiology and prognosis

Diseases of the gallbladder are widespread, affecting approximately 10–15% of the adult population. 26 Clinical surveys conducted in Europe, North and South America and Asia indicate that the prevalence rates for gallstone disease range from 5.9% to 25%27–37 and tend to increase with age. A clinical ultrasound survey conducted in the UK reported prevalence rates of 12% among men and 22% among women over 60 years of age. 35 A multicentre, population-based study conducted in Italy has reported a cumulative incidence of gallstone disease of 0.67% per year (0.66% in men and 0.81% in women). 38 In the developed world, gallstone disease is the most common gastrointestinal complaint for which people are admitted to hospital. 2,10,39 Hospital admissions increased in England by 30% in men and by 64% in women from 1989–90 to 1999–2000. 3

Natural history studies have shown low mortality from gallstone disease, with typically < 1% of people dying from gallbladder-related causes. 3,12,13,17 The natural course of gallstone disease is benign, with a relatively low progression from asymptomatic disease to symptomatic disease. In a recent population-based study, the overall frequency of symptom development in asymptomatic people was around 20% over a long follow-up period (mean 8.7 years). 12 An early natural history study (in which people were found to have gallstones through routine screening), showed that 10%, 15% and 18% of people assessed became symptomatic at 5, 10 and 15 years, respectively. 40

Overall, the annual risk of developing complications for people with asymptomatic gallstones is low, about 0.1–0.2%. In contrast, in people with symptomatic gallstone disease, the annual rates of developing complications have been reported to be higher, 1–3%. 41–43 The Italian Group for the Epidemiology and Prevention of Cholelithiasis study reported an annual incidence of complications of 0.3% for asymptomatic people and 0.7% for symptomatic people. 13 In particular, there is a proportion of people with uncomplicated gallstone disease who may experience only a few episodes of biliary pain without developing serious symptoms or complications for many years. 44,45 A recent multicentre study found that a considerable proportion of people with mild and severe symptoms (58% and 52%, respectively) did not experience subsequent episodes of biliary pain during the 10-year follow-up period and 10% suffered from gallstone complications, indicating a benign course of the disease even in those with initially severe symptoms. 12 Moreover, the severity of the disease does not seem to increase over time. 21,46

In the UK and in North America, the number of surgical procedures for gallstone disease increased steadily between the 1950s and 1990s, reflecting both the rise in prevalence of gallstone disease and the use of cholecystectomy as the treatment of choice. 3,47 Rates of surgical procedures stabilised in both countries towards the end of the twentieth century. 3,48

Impact of health problem

From a patient perspective, the defining symptom of gallstone disease is pain. 49,50 Commonly, general abdominal symptoms intensify over a period of time and become regular pain attacks, which require medical attention. Although the pain experienced may be described as ‘non-specific’,7,49,50 it tends to occur in a defined location and with characteristic patterns as described above. Best medical therapy includes the prescription of analgesics and, when necessary, antibiotics.

The most common complications associated with gallstones are acute cholecystitis, common bile duct (CBD) stones and acute pancreatitis. Acute cholecystitis is caused by a gallstone obstructing the cystic duct and results in unresolving upper right quadrant pain, nausea, vomiting, anorexia and fever. 51 CBD stones are found in up to 15% of people who undergo cholecystectomy. They may be asymptomatic or accompanied by biliary pain, jaundice, pancreatitis or cholangitis. 52 CBD stones can cause acute pancreatitis by obstructing the main pancreatic duct. 53 The typical symptoms of gallstone-related pancreatitis are epigastric abdominal pain, nausea and vomiting. 51 Pancreatitis is severe in over 20% of people with gallstone disease and can cause death in about one-third. 54

Even though removal of the gallbladder is considered the standard treatment for symptomatic gallstones, it does not guarantee eradication of symptoms. 55 Up to approximately 40% of people may continue to experience pain and abdominal symptoms after surgery. 56 In particular, marked biliary pain has been described in 4–9% of people after cholecystectomy, while persistent abdominal pain or non-specific pain persists in about 13–37% of people. 57–62 A recent systematic review of the literature found that up to one-third of people suffered continuing pain after cholecystectomy and up to 14% of people experienced de novo pain. 63 Some investigators have also reported a persistent pain similar to that experienced pre-operatively in about 20% of people with gallstones. In a prospective study conducted in Denmark, 21% of people experienced the same type of pain after surgery. 64 Similarly, in a RCT conducted in the UK, 19% of people complained of biliary pain 5 years after open cholecystectomy. 65 No difference has been observed between open and laparoscopic surgery in terms of persistent pain. 66,67 The term ‘post-cholecystectomy syndrome’ is an umbrella term that has been widely used to describe, though not accurately, the range of symptoms which occur after cholecystectomy. 68 The term ‘persistent post-cholecystectomy symptoms’ has been suggested as a more accurate description of these symptoms. 69 These symptoms include biliary and non-biliary abdominal pain, gastrointestinal disorders, dyspepsia, heartburn, nausea, vomiting, jaundice and cholangitis. Severe symptoms which occur early after surgery may represent complications of cholecystectomy, whereas those that manifest later (after months or years) are probably unrelated to cholecystectomy and explained by non-biliary causes. Non-biliary causes are more likely if the symptoms are similar to those experienced pre-operatively and no stones are found in the gallbladder. Recent research has suggested that, in some people, functional gastrointestinal disorders and not gallstone disease may be the cause of persistent post-surgery symptoms. 16 Nevertheless, there is no consistent pathophysiological explanation for persistent post-cholecystectomy symptoms, and in about 5% of people the reason for persistent abdominal pain remains unknown. 70

Management of disease

Management of gallstone disease may include pharmacological, non-pharmacological and surgical interventions. Surgery is the definitive way to treat gallstone disease. Laparoscopic cholecystectomy is currently preferred over open cholecystectomy for elective surgery in symptomatic gallstones. 5,71 In the UK, people with biliary pain are commonly put on a waiting list and undergo elective surgery several months after the original clinical diagnosis,72 although early and urgent surgery during the same admission is becoming more common.

People with symptomatic gallstones who are unfit or unwilling to undergo surgery are managed conservatively or may be offered alternative non-surgical treatments. In particular, conservative management is a feasible and relatively safe option for pregnant women and elderly people who are less likely to tolerate surgery. 73–75 The terms ‘observation’ and ‘conservative management’ are not clearly defined in the current literature and appear to be used interchangeably. In general, observation/conservative management in the context of gallstone disease involves the prescription of analgesics to relieve the biliary pain. Typical therapy includes narcotic analgesics (e.g. opiates) or non-steroidal anti-inflammatory drugs (e.g. ibuprofen) together with generic lifestyle advice. 15,55,76–78 Complicated biliary pain (i.e. inflammation of the gallbladder, namely acute cholecystitis) usually requires additional therapy with antibiotics. 15

Medical therapy for gallstone disease using bile acids, for example ursodeoxycholic acid, to dissolve stones (especially cholesterol stones) is presently restricted to a small, highly selected group of symptomatic people for whom a surgical intervention is not recommended and recurrence is likely to have particularly adverse consequences. 15 Novel experimental animal models and preliminary clinical findings indicate that future research could focus on the mechanisms of intestinal absorption of cholesterol, on hepatic cholesterol biosynthesis as well as on the role of gallstones genes. 15,79

Other non-pharmacological, non-surgical treatments include dissolution/fragmentation of gallstones using the cholesterol solvent methyl tert-butyl ether (MTBE) and extracorporeal shock wave lithotripsy (ESWL). These treatment options are very rarely used in clinical practice because of their potential side effects (MTBE), high recurrence rate in people with multiple stones (ESWL) and occurrence of transient biliary pain after successful stone fragmentation (ESWL). 5,80,81

The widespread use of laparoscopic cholecystectomy together with the use of imaging techniques for a more accurate diagnosis of symptomatic gallstone disease has contributed to the decline in non-surgical medical interventions such as oral bile acids and ESWL. 80–82

Current service cost

Even though the majority of gallstones are asymptomatic and do not require treatment, gallstone disease is still the most expensive of the digestive disorders in industrialised societies. 83 Considerable resources are involved in managing people with symptomatic gallstones. The bulk of the economic burden is mainly due to the costs associated with surgery for the removal of the gallbladder. 2,3

According to the NHS 2011/12 tariff for admitted patient care and outpatients procedures, for laparoscopic cholecystectomy, the day case tariff was £1689, whereas the elective spell tariff was £1370 and the non-elective spell tariff £3197. For open cholecystectomy, the combined day case and spell tariff was £2285, whereas the non-elective spell tariff was £4513. 84 Best practice tariffs were introduced in England in 2010–11 to incentivise cholecystectomy on a day-case basis, where clinically appropriate. 85

Variation in services and/or uncertainty about best practice

In the UK, people with biliary pain are usually put on a waiting list and undergo elective surgery several months after establishment of a clinical diagnosis of gallstone disease. The waiting time for elective cholecystectomy varies from hospital to hospital depending on the resources available, but generally lies between 4 and 12 months (although current clinical targets have led to a shortening of waiting lists). 86–89 There is a marked variation between UK health trusts in terms of rates for laparoscopic cholecystectomy. Approximately half of trusts perform < 5% of cholecystectomies as day cases, with many trusts performing no day procedures at all. 90 The variation is likely to be explained by the local resources available in each trust and the number of people with symptomatic gallstone disease being treated during the same emergency admission or electively after undergoing a period of conservative treatment.

Relevant national guidelines, including National Service Frameworks

There are no current published national guidelines on the management of gallstone disease.

Summary of the surgical procedures under assessment

Laparoscopic cholecystectomy is the current standard surgical procedure for the management of gallstone disease. Since its development in the late 1980s,91 laparoscopic cholecystectomy has increasingly replaced open cholecystectomy, which remained the gold standard surgical treatment for over 100 years. 92 Laparoscopic cholecystectomy has been demonstrated to be a safe and cost-effective procedure with similar mortality and complications rates to those of open cholecystectomy, but significant shorter hospital stay, quicker recovery93 and lower total cost. 94

Current usage in the National Health Service

At present, the definitive treatment choice for gallstone disease is surgery. Cholecystectomy is one of the most common surgical procedures performed in the UK. About 70,000 cholecystectomies were performed in England between 2011 and 2012. 107 The majority of these surgical procedures were performed after an elective admission and undertaken laparoscopically. 90 Laparoscopic cholecystectomy is technically more challenging to perform than the traditional open cholecystectomy and the success of the operation depends on the experience, specialisation and technical skills of the surgeon who performs the procedure. 5 In Scotland, the total cholecystectomy rate increased by 18.7% between 1989 and 1993. 108 Hospital Episode Statistics data for England for 2011–12 reported 65,926 admissions for excision of the gallbladder with a mean waiting time of 81 days. There were 6148 (9%) emergency admissions and 57,920 (88%) on the waiting list. 107

While waiting for surgery, people with symptomatic gallstones are treated conservatively. People who are unfit or unwilling to undergo surgery are managed by means of a ‘wait and see’ therapeutic strategy. Similarly, people with a single episode of biliary pain and no complications of gallstone disease may be potential candidates for conservative management, with a reasonable chance to remain symptom free.

Population

The population considered for this assessment is adults with symptomatic uncomplicated gallstone disease (biliary pain or acute cholecystitis) who are examined in a secondary care setting and considered suitable for cholecystectomy.

Clinical diagnosis of gallstone disease is usually confirmed by imaging and laboratory tests. Transabdominal ultrasonography is the standard imaging technique for the diagnosis of gallbladder stones. The technique is accurate, non-invasive and very widely available. 7,109,110

Intervention: cholecystectomy

Cholecystectomy is the standard treatment for symptomatic gallstone disease and its frequency is increasing worldwide. 50 Nowadays, nearly all cholecystectomies in the UK are performed using a minimally invasive laparoscopic approach (laparoscopic cholecystectomy). 5,111 Any cholecystectomies performed openly or laparoscopically (or variants thereof) are considered suitable for inclusion.

Comparator: observation/conservative management

The comparator intervention considered in this assessment is observation/conservative management. ‘Conservative management’ in the context of gallstone disease typically comprises the prescription of analgesics/anti-inflammatory drugs along with lifestyle advice. People with cholecystitis and signs of inflammation are usually prescribed antibiotics.

‘Delayed surgery’ is not considered a suitable comparator. The focus of recently published randomised trials and meta-analyses looking at early compared with delayed cholecystectomy (open and laparoscopic) has been on the optimal timing of surgical intervention as well as operative outcomes (in a patient population scheduled to receive surgery) rather than on the necessity of cholecystectomy in people with uncomplicated symptomatic gallstones presenting to secondary care.

Types of studies

For assessing the clinical effectiveness of cholecystectomy compared with observation/conservative management in people with biliary pain or acute cholecystitis, the following types of study design were deemed suitable for inclusion:

• RCTs which randomised people to either cholecystectomy (open or laparoscopic) or observation/conservative management, irrespective of study language, blinding and publication status

• non-randomised comparative studies in which people received either cholecystectomy (open or laparoscopic) or observation/conservative management.

The following types of report were excluded:

• reviews, editorials and opinions

• case series and case reports

• reports published in non-English languages for which a translation could not be organised.

Types of participants

Adults (18 years and older or ‘adults’ as defined by the triallists) with symptomatic gallstone disease (biliary pain or acute cholecystitis) being considered for cholecystectomy in a secondary care setting for the first time. Gallstone-related disease needed to be confirmed by ultrasonography. For the purpose of this assessment, ‘first episode of symptomatic gallstones’ was defined as the first instance in which the patient presented to secondary care attention (with the possibility of receiving surgical treatment) even though the clinical history documented previous clinical symptoms or pain attacks. This criterion was later relaxed to allow inclusion of studies in which up to 25% of participants had previously presented to secondary care for symptomatic gallstones. First episode was assumed unless studies explicitly described cases as recurrent. Eligibility for cholecystectomy, if not explicitly reported, was assumed in RCTs whereas non-randomised studies were assessed on an individual basis on this criterion. An accepted definition of ‘acute cholecystitis’ is based on a combination of relevant clinical symptoms (e.g. pain localised to the right upper quadrant of the abdomen; temperature exceeding 37.5 °C, leucocytosis greater than 10 × 10⁹/l, increased C-reactive protein level) and ultrasonographic evidence of gallstones.

People with acute severe cholecystitis (e.g. obstruction of the cystic duct or neck of the gallbladder by gallstones) and/or cholangitis (inflammation, usually infection, of a bile duct) or pancreatitis were not considered suitable for inclusion, as they normally require urgent or emergency intervention. Similarly, people with symptomatic gallstones complicated by severe concomitant diseases, critically ill people, and/or people who were judged to be unfit or unsuitable for surgery (e.g. pregnant women) were not considered within the scope of this review.

Therefore, for those with acute cholecystitis, we included any study population presenting with acute, uncomplicated, cholecystitis.

Intervention

The intervention considered was surgical removal of the gallbladder (laparoscopic or open cholecystectomy).

Comparator interventions

The comparator interventions considered were observation (watchful waiting) and/or conservative treatment. Conservative treatment refers here to a course of analgesics/anti-inflammatory drugs accompanied by lifestyle advice. People with cholecystitis and signs of inflammation may also be prescribed antibiotics.

Types of outcomes

The following types of outcome measure were considered:

• disease-related morbidity

  • recurrence of symptoms

  • complications (e.g. pancreatitis)

  • number of visits to primary care settings or hospital emergency department

  • analgesic requirements

  • need for surgical, endoscopic or radiological intervention

  • need for further medical intervention

  • mortality

• surgery-related morbidity

  • bile duct injury

  • infection/bleeding

  • reoperation rate

  • diarrhoea

  • recurrent pain

  • mortality

• patient-driven outcomes

  • generic and disease-specific quality of life (QoL; as defined by the studies’ authors)

• cost of initial and any subsequent treatments.

Number and type of studies included

The primary literature searches identified 6779 potentially relevant citations. Seventy-seven reports were selected for full-text assessment. Seventy-one reports were subsequently excluded (see Appendix 5). Two RCTs published in six reports were included in this assessment. 46,55,77,78,113,114 No eligible non-randomised comparative studies were identified by the literature searches. A flow diagram of the screening process is outlined in Figure 3.

FIGURE 3.

Flow chart for the identification and selection of studies.

Appendix 5 provides the bibliographic details of the included and excluded studies.

Number and type of studies excluded

A list of the 71 full-text papers that were excluded along with the reasons for their exclusion is given in Appendix 5. These reports were excluded because they failed to meet one or more of the inclusion criteria in terms of the type of study, participants, intervention/comparator or outcomes reported.

Characteristics of the included studies

The search identified two randomised trials, Schmidt et al. 113 and Schmidt et al. ,46 published in full in six reports. 46,55,77,78,113,114 Participants were enrolled consecutively in both trials and data were collected prospectively. Both trials were conducted in Norway.

Table 1 presents the demographic information from the studies. The two trials included a total of 201 participants at enrolment. There were no dropouts during the study period and all 201 randomised participants were included in the statistical analyses. Participants’ diagnosis was uncomplicated symptomatic gallstones (biliary pain only) in Schmidt et al. 113 and acute cholecystitis in Schmidt et al. 46 In total, 149 women and 52 men were assessed. Of these, 72 women and 27 men were randomised to surgery, while 77 women and 25 men were randomised to observation. Median age was 50 years (range 20–79 years) in Schmidt et al. 113 and 58 years (range 27–77 years) in Schmidt et al. 46 Both trials enrolled some participants who had previously presented to secondary care for gallstone disease: 30 participants (22%) in Schmidt et al. 113 and 11 participants (17%) in Schmidt et al. 46 Both trials also reported the number of participants with concomitant diseases, specifically heart disease, diabetes and/or obstructive lung disease, in total 16 participants (12%) in Schmidt et al. 113 and 12 participants (19%) in Schmidt et al. 46 Baseline QoL, as assessed by the psychological general well being index (PGWB), Nottingham Health Profile (NHP) part II, pain score and Visual Analogue Pain Score (VAPS), were comparable across the studies. The mean follow-up was 14 years (range 13–16 years) for both trials. Similarly, the median follow-up was 14 years in Schmidt et al. 46

Table 2 presents the key features of the two included studies.

Risk of bias of the included studies

The two published trials were assessed using The Cochrane Collaboration’s risk of bias tool. 112 Tables 3 and 4 and Figures 4 and 5 present the summary findings of this assessment. Both trials were judged to be at low risk for all domains assessed by the tool.

FIGURE 4.

Risk of bias graph for both included studies.

FIGURE 5.

Risk of bias summary figure.

Number of participants undergoing surgery (14 years)

A total of 46 out of 102 (45%) participants randomised to observation and 87 out of 99 (88%) participants randomised to surgery underwent cholecystectomy (Figure 6). This was a statistically significant difference between the groups (risk ratio = 0.50; 95% CI 0.34 to 0.73; p = 0.0004).

FIGURE 6.

Forest plot of participants having surgery (14 years).

Pain attacks (14 years)

A total of 26 out of 102 (25%) participants randomised to observation and 12 out of 99 (12%) participants randomised to surgery experienced pain attacks (Figure 7). The difference between the groups was not statistically significant (risk ratio = 1.62; 95% CI 0.43 to 6.17; p = 0.48).

FIGURE 7.

Forest plot of pain attacks (14 years).

In general, pain attacks were reported more frequently in participants with uncomplicated symptomatic gallstone disease managed conservatively. One-third (33%) of participants with uncomplicated symptomatic gallstones (biliary pain only) randomised to observation experienced pain attacks after randomisation compared with only 9% of participants with acute cholecystitis. It is worth pointing out that a similar proportion of participants randomised to surgery experienced post-randomisation pain attacks across the groups (12%, Schmidt et al. ;113 13%, Schmidt et al. 46). Significantly more participants with uncomplicated symptomatic gallstones (biliary pain only) randomised to observation experienced pain attacks than those randomised to surgery (χ² = 9.10; p = 0.0026), whereas no difference was observed between intervention groups within participants with an initial diagnosis of acute cholecystitis (χ² = 0.24; p = 0.6253).

Gallstone-related complications: total (14 years)

Significantly more participants randomised to observation (14/102; 14%) than those randomised to surgery (2/99; 2%) experienced any of the specified gallstone-related complications (risk ratio = 6.69; 95% CI 1.57 to 28.51; p = 0.01) (Figure 8).

FIGURE 8.

Forest plot of total gallstone-related complications (14 years).

Details of participants with acute cholecystitis, CBD stones and acute pancreatitis are presented below.

Gallstone-related complications: acute cholecystitis (14 years)

Participants randomised to observation were more likely to suffer from acute cholecystitis as a complication (Figure 9). Nine of the 102 (9%) participants randomised to observation and none of the 99 (0%) participants randomised to surgery experienced acute cholecystitis as a complication. The difference between intervention groups was statistically significant (risk ratio = 9.55; 95% CI 1.25 to 73.27; p = 0.03).

FIGURE 9.

Forest plot of episodes of acute cholecystitis (14 years).

It is worth noting that, among those treated conservatively, new complicated events occurred more frequently in those in whom the initial diagnosis was acute cholecystitis than in those initially diagnosed as having uncomplicated symptomatic gallstones (biliary pain only).

Gallstone-related complications: common bile duct stones (14 years)

Few participants in both intervention groups suffered from CBD stones at the 14-year follow-up assessment (Figure 10). Four of the 102 (4%) participants randomised to observation and 1 of the 99 (1%) participants randomised to surgery experienced CBD stones. The difference between the groups was not statistically significant (risk ratio = 2.86; 95% CI 0.47 to 17.59; p = 0.26).

FIGURE 10.

Forest plot of episodes of CBD stones (14 years).

Gallstone-related complications: acute pancreatitis (14 years)

The risk of experiencing acute pancreatitis was the same in both intervention groups (Figure 11). One of the 102 (1%) participants randomised to observation and 1 of the 99 (1%) participants randomised to surgery experienced acute pancreatitis (risk ratio = 0.99; 95% CI 0.06 to 15.44; p = 0.99). Both events occurred in participants with uncomplicated symptomatic gallstones (biliary pain only) whereas participants with acute cholecystitis did not experience this complication.

FIGURE 11.

Forest plot of episodes of acute pancreatitis (14 years).

Surgery-related complications: total (5 years)

As expected, the risk of surgery-related complications was significantly lower in the observation group (Figure 12). In total, 7 out of 102 (7%) participants randomised to observation and 19 out of 99 (19%) participants randomised to surgery experienced surgery-related complications (risk ratio = 0.36; 95% CI 0.16 to 0.81; p = 0.01).

FIGURE 12.

Forest plot of total surgery-related complications (5 years).

Further details of individual surgery-related complications are presented below.

Surgery-related complications: intra-abdominal infection/bile leakage (5 years)

The risk of suffering from intra-abdominal infection/bile leakage was similar in both intervention groups (Figure 13). A total of 4 out of 102 (4%) participants randomised to observation and 3 out of 99 (3%) participants randomised to surgery experienced intra-abdominal infection or bile leakage (risk ratio = 1.30; 95% CI 0.30 to 5.63; p = 0.73). Five of the seven events were experienced by participants originally diagnosed with uncomplicated symptomatic gallstones (biliary pain only).

FIGURE 13.

Forest plot of intra-abdominal complications – infection/bile leakage (5 years).

Surgery-related complications: wound infection/dehiscence (5 years)

The number of participants who suffered from wound infection/dehiscence after surgery was the same in both intervention groups (Figure 14). One of the 102 (1%) participants randomised to observation and 1 of the 99 (1%) participants randomised to surgery experienced wound infection or dehiscence (risk ratio = 0.99; 95% CI 0.06 to 15.44; p = 0.99). Both events occurred in participants diagnosed with uncomplicated symptomatic gallstones (biliary pain only) whereas no participants diagnosed with acute cholecystitis experienced this complication.

FIGURE 14.

Forest plot of episodes of wound infection/dehiscence (5 years).

Surgery-related complications: bile duct injury

Figure 15 shows that only one participant randomised to surgery suffered from bile duct injury within 5 years (risk ratio = 0.31; 95% CI 0.01 to 7.42; p = 0.47).

FIGURE 15.

Forest plot of bile duct injury cases (5 years).

Surgery-related complications: reoperation (5 years)

The number of participants who required reoperation was the same in both intervention groups (Figure 16). One of the 102 (1%) participants randomised to observation and 1 of the 99 (1%) participants randomised to surgery underwent reoperation (risk ratio = 0.99; 95% CI 0.06 to 15.44; p = 0.99). Both participants who underwent reoperation were originally diagnosed with uncomplicated gallstones (biliary pain only).

FIGURE 16.

Forest plot of reoperation cases (5 years).

Surgery-related complications: minor complications (5 years)

Significantly fewer participants randomised to observation experienced minor complications (1/102; 1%) than participants randomised to surgery (13/99; 13%) (risk ratio = 0.11; 95% CI 0.02 to 0.56; p = 0.008) (Figure 17). A slightly larger proportion of participants originally diagnosed with acute cholecystitis experienced minor complications (8/64; 13%) than participants with uncomplicated symptomatic gallstones (biliary pain only) (6/137; 4%). The nature of minor complications was not described in the included trials.

FIGURE 17.

Forest plot of minor complications (5 years).

Admission due to gallstone-related pain (5 years)

A total of 16 out of 102 (16%) participants randomised to observation and 5 out of 99 (5%) participants randomised to surgery were admitted to hospital because of gallstone-related pain (Figure 18). There was no statistically significant difference between the intervention groups at the 5-year follow-up (risk ratio = 2.69; 95% CI 0.57 to 12.68; p = 0.21).

FIGURE 18.

Forest plot of number of admissions due to gallstone-related pain (5 years).

For both uncomplicated symptomatic gallstones (biliary pain only) and acute cholecystitis groups, more participants randomised to observation than to surgery were subsequently admitted to hospital for gallstone-related pain (12 vs. 2 people with biliary pain and 4 vs. 3 people with acute cholecystitis). It is worth noting that the difference between intervention groups was statistically significant for participants with uncomplicated symptomatic gallstones (biliary pain only) (χ² = 7.79; p = 0.0052) but not for those with acute cholecystitis (χ² = 0.10; p = 0.7542).

Mortality rate (14 years)

Schmidt et al. 113 reported that a total of 8 out of 137 participants with uncomplicated symptomatic gallstones (biliary pain only) had died by the 5-year follow-up, but they did not provide separate data for each intervention group. At the 14-year follow-up assessment, 8 out of 68 participants randomised to surgery and 11 out of 69 participants randomised to observation had died.

Schmidt et al. 46 reported that 4 out of 31 participants with acute cholecystitis who were randomised to surgery had died by the 5-year follow-up, whereas no deaths were observed among participants randomised to observation. By the 14-year follow-up, 8 out of 31 participants randomised to surgery and 10 out of 33 participants randomised to observation had died. None of the deaths was caused by gallstone disease.

A total of 16 out of 102 (16%) participants randomised to observation and 21 out of 99 (21%) participants randomised to surgery died (Figure 19). The difference between the groups was not statistically significant (risk ratio = 0.73; 95% CI 0.41 to 1.31; p = 0.30).

FIGURE 19.

Forest plot of number of deaths (14 years).

Further surgical intervention required

Schmidt et al. 113 reported that, at the 14-year follow-up, 4 out of 60 (7%) participants with uncomplicated symptomatic gallstones (biliary pain only) randomised to surgery who did undergo cholecystectomy subsequently required endoscopic retrograde cholangiopancreatography. Common bile duct stones were detected in one of these four participants.

Median time to surgery (from randomisation)

Schmidt et al. 113 reported that the median time to surgery for participants with uncomplicated symptomatic gallstones (biliary pain only) was 3 months (range 0–68 months) for participants randomised to surgery and 28 months for participants randomised to observation (no range was given for the observation group). Schmidt et al. 46 stated that patients who were randomised to surgery, on average, underwent an operation 4 months (median) after randomisation (range 1–13 months). No details were provided for those patients randomised to observation who eventually underwent surgery.

Quality of life

We were not able to include QoL measures in the statistical analyses as measures of variability were not reported and were not available from the trial investigators. Quality-of-life measures were assessed only at 5 years. In the paragraphs below, scores for the observation group are reported first, followed by scores for the surgery group. None of the differences between groups was statistically significant.

Quality of life: Psychological General Well Being Index

In the Schmidt et al. trial113 both intervention groups reported PGWB scores lower than 105 at 6 months, which is the score normally expected in healthy control subjects (mean score of 99.9 for the observation group and 103.3 for the surgery group, respectively). 113 This indicates a slightly poorer level of general well-being than in healthy control subjects. In addition, the observation group reported lower scores than the surgery group. This difference was, however, not statistically significant. Notably, at this time point, 7 of the 35 participants in the observation group who crossed over to surgery had actually been operated on.

At 6 months, 46 participants with acute cholecystitis reported higher scores and, therefore, greater well-being than the average score of 105 expected in healthy control subjects (mean score of 110.2 for the observation group and 106.2 for the surgery group). In contrast to the Schmidt et al. trial,113 participants in the observation group reported higher (but not significantly higher) scores than participants who underwent surgery.

At 12 months, participants in both trials reported levels of well-being lower than those expected of healthy control subjects. In the Schmidt et al. trial,113 the mean PGWB scores were 101.8 for the observation group and 102.3 for the surgery group. Similarly, in the Schmidt et al. trial,46 the mean PGWB scores were 103.1 for the observation group and 103.2 for the surgery group. It is worth noting, however, that, at this time point, participants’ scores in both trials were similar.

In both trials, participants managed conservatively reported higher well-being scores at 60 months than those who underwent surgery. In the Schmidt et al. trial,113 the mean PGWB score was 104.1 for the observation group compared with 101.6 for the surgery group, whereas in the Schmidt et al. trial46 the mean PGWB score was 112.0 for the observation group and 102.5 for the surgery group. In Schmidt et al. participants randomised to observation reported levels of general well-being greater than scores which would be expected in healthy control subjects. 115

Quality of life: Nottingham Health Profile part II

At all time points, both trials reported NHP scores of < 2, indicating a limited number of specified domains (i.e. job, housework, social life, home life, sex life, interests/hobbies, holidays) affected by gallstone disease. In particular, the mean NPH scores for the observation group and the surgery group were, respectively, 1.6 and 1.1,113 and 1.1 and 1.246 at 6 months; 1.6 and 1.0113 and 0.9 and 1.246 at 12 months; and 1.3 and 1.9113 and 0.7 and 1.446 at 60 months. In both trials, scores did not vary significantly over time or between intervention groups.

Quality of life: pain score

Pain scores among participants with uncomplicated symptomatic gallstones (biliary pain only)113 showed a tendency to decrease between 6 and 60 months, with all scores being at the lower end of the possible range of 0–16. At 6 months, participants in the observation group had a mean score of 4.1 and participants in the surgery group a mean score of 2.2. At 12 months, the mean score was 3.7 and 2.3 for the observation group and the surgery group, respectively, whereas at 60 months the scores were 2.4 and 2.0, respectively. Pain scores were consistently higher for the observation group than for the surgery group (albeit the difference was not statistically significant).

Mean scores among participants with acute cholecystitis46 were also generally at the low end of the range (2.1 and 2.0 at 6 months, 3.1 and 2.4 at 12 months and 1.3 and 2.6 at 60 months for the observation group and surgery group, respectively). Scores for the surgery group increased between 6 and 60 months while scores for the observation group increased from 6–12 months and then decreased between 12 and 60 months. Furthermore, at 60 months, pain score for the surgery group was greater than that of the observation group (the only time point at which this occurred in either of the two trials).

Quality of life: Visual Analogue Pain Score

Schmidt et al. 113 reported mean VAPS at the lower end of the possible range (0–100) among participants with biliary pain only (14.7 and 9.4 at 6 months, 15.7 and 7.0 at 12 months and 11.5 and 5.5 at 60 months for the observation group and surgery group, respectively). In particular, scores in the observation group were consistently higher (but not significantly higher) than those in the surgery group. Pain scores tended to decrease in the surgery group between 6 and 60 months, whereas in the observation group scores increased between 6 and 12 months and subsequently decreased to pre-12-month levels at 60 months.

Mean VAPS in participants with acute cholecystitis allocated to surgery46 increased over time while in participants managed conservatively it varied over time, with a similar trend to that observed in the Schmidt et al. trial (8.1 and 5.4 at 6 months, 15.1 and 9.9 at 12 months and 6.2 and 11.3 at 60 months for the observation group and surgery group, respectively). 113

Model development

A de novo economic model was developed to compare alternative treatment strategies for people with symptomatic gallstone disease (biliary pain or cholecystitis). According to the purpose of this assessment two strategies were considered:

• conservative management

• surgical management (cholecystectomy).

The structure of the model was informed by the findings of the trials included in the systematic review of clinical effectiveness, other existing evidence, as well as by the advice from health-care professionals within the research team for this assessment. The perspective adopted for the analysis is that of the UK NHS and the Personal Social Services.

The assumed care pathway of the model

The model described in this chapter is based on two clinically relevant care pathways for gallstone disease: a surgery care pathway and a conservative management care pathway (Figures 25 and 26). Both pathways were derived from the care pathway reported in Chapter 2 (Definition of the decision problem). As the current standard treatment for gallstone disease in the UK is cholecystectomy, the conservative management pathway portrays a plausible pathway that people with gallstones can follow.

FIGURE 25.

Surgery care pathway. Rectangles represent treatment options; ovals represent health states; arched arrows above the ovals represent health states in which people may return; and arrows indicate the possible directions in which individuals could move at the end of each cycle.

FIGURE 26.

Conservative management care pathway. Rectangles represent treatment options; ovals represent health states; arched arrows above the ovals represent health states in which people may return; and arrows indicate the possible directions in which individuals could move at the end of each cycle.

The first care pathway (Figure 25) considers patients who are treated with surgery (mainly laparoscopic cholecystectomy). There are five health states that could arise from the surgical management: peri- and post-operative complications, no symptoms, post-surgical symptoms (such as pain that persists after surgery) and death. Patients who survive surgery can either have no complications or suffer from peri- or post-operative complications. These complications can either be managed without any further surgical intervention or require further surgery (e.g. surgery for bile duct injury). Those patients who experience no complications could end up with symptoms or they could have no symptoms. In subsequent cycles there is always a chance that patients die when they are in either a symptom or no symptom state from all-cause age-related mortality.

The second pathway (Figure 26) considers three health states for patients with gallstone disease who are managed conservatively: no symptoms, symptoms or death. Patients who are free from symptoms after conservative treatment have a chance of staying in this state (no symptoms) or they may continue to experience the recurrence of symptoms over time and, therefore, move into a symptoms state. Patients who continue to experience symptoms have a chance of continuing to be treated conservatively or they can be offered surgery. As in the surgery pathway, patients always have a chance of dying from all-cause age-related mortality. The key outcome of the conservative management (number of patients without symptoms) will be influenced by whether or not the treatment resolves the symptoms and, therefore, whether or not patients seek any further medical intervention, in particular surgery.

Model structure

A Markov model was developed to model the alternative management strategies (i.e. conservative management or surgery) that patients with symptomatic gallstone disease may be offered over time and to provide estimates of costs, clinical effectiveness [measured in quality-adjusted life-years (QALYs)] and cost-effectiveness (Figure 27). The model estimates the costs and consequences of both interventions and of any further events – such as the treatment for recurrent symptoms and/or complications. The model was developed using TreeAge Pro 2012 (TreeAge Software, Inc., Williamstown, MA, USA).

FIGURE 27.

Markov model structure.

The model was based on a hypothetical cohort of patients presenting with symptomatic gallstone disease (biliary pain or cholecystitis). In the model, symptomatic patients receive an initial treatment of either conservative management or surgery (cholecystectomy) and then move into different health states, either surviving or dying. The assumption that over a given period of time a patient could die is always built into the model. Death takes into account that patients suffering from gallstone disease (i) are exposed to a very small risk of death when they undergo surgery and (ii) have a chance of dying from all other causes at any point in time (this chance is assumed to be equivalent to that of all-cause age-related mortality). 118 Death is referred to as an absorbing state (an absorbing state in a Markov model is a state that once entered cannot be exited again). Those patients who survive cholecystectomy end up in one of two health states: they can either have peri- or post-operative complications or they can have no complications for the remainder of the cycle. Those patients who have post-operative complications (e.g. bile duct injury, port incisional hernia) remain in the complications post-surgery state for the remainder of the cycle.

Similarly, those patients who have conservative management end up in one of two health states: their symptoms can either be resolved or they can persist continuously or intermittently over time. It is assumed that no further treatment is sought by those whose symptoms resolve. Those individuals who continue to experience symptoms can either undergo surgery or they can continue to be treated conservatively. In the model it is assumed that all individuals who continue to experience symptoms will present for surgery within 5 years, so by the end of 5 years all the people left under conservative management are symptom free.

The following are the key assumptions of the economic modelling exercise:

• The time horizon of the model is 5 years. This time horizon was chosen as long-term follow-up data46 indicate that the majority of people with symptomatic gallstones undergo cholecystectomy during the first 5 years.

• The cycle length (i.e. the time period by which transitions between health states are modelled) is 1 year.

• In the base case, the typical person with symptomatic gallstone disease entering the model is a female of 51 years of age. This choice is justified by the average age at which people currently undergo surgery to remove their gallbladder and by the fact that women are more likely to suffer from gallstone disease than men. 84

The results of the model are presented in terms of incremental cost per QALY. The costs and benefits incurred beyond 1 year were discounted at 3.5% per year in accordance with the current National Institute for Health and Care Excellence (NICE) guidelines. 119

Data requirements and model inputs

Complications during and after laparoscopic surgery

According to the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES) the current rate of major bile duct injury in laparoscopic cholecystectomy has stabilised at 0.1–0.6%120–126 and series with no major bile duct injuries have been reported. 127 Although many believe that the rate of major bile duct injury in open cholecystectomy is lower than laparoscopic cholecystectomy, controversy remains. 125 Post-operative complications include bile leak, retained stone, wound infection, pneumonia and bile duct injury. The estimate used in the model for the probability of having peri-operative complications was 0.1 (10%), inclusive of those patients who had their laparoscopic cholecystectomy converted into open cholecystectomy. This value was based on advice from clinical experts and is a conservative assumption (in the sense that it is likely to be a lower rate and, based on SAGES data; therefore, our estimates of cost-effectiveness of surgery are less favourable than they would have been if SAGES data had been used). An estimate for the proportion of patients whose complications require further surgery was based on the information available to patients who undergo cholecystectomy in the NHS in England and expert opinion. 128 These values vary from 0.01 to 0.002 (1/100 to 1/500) and the estimate used in the base analysis was 0.01 (1/100).

Symptoms after surgery

Although cholecystectomy is the preferred treatment for symptomatic gallstones, the relationship between cholecystectomy and post-cholecystectomy symptoms is still unclear. 63 Individuals with gallstones have various abdominal symptoms that may be caused by gallstones or may be present as a separate condition but with common physiology. 50 The accompanying abdominal symptoms are referred to as functional gastrointestinal disorders. As both functional disorders and gallstones manifest with the same clinical symptoms (pain and discomfort) it is difficult to distinguish between the two conditions. Lack of clear distinction makes it difficult to treat the symptoms if they remain after the operation. Existing literature suggests that a proportion of people may still suffer from persistent or de novo symptoms after surgery. 63,129–131 Post-cholecystectomy syndrome occurs when abdominal symptoms persist after cholecystectomy. Reports suggest that about 4–40% of individuals who have their gallbladder removed may continue to experience symptoms after surgery. 57–62,132 The probability of having symptoms after laparoscopic cholecystectomy was informed by existing evidence and advice from clinical experts. The value used in the base-case analysis was 0.1 (10%).

Symptoms and surgery after conservative management

For people with gallstones treated conservatively, the model utilises the probability of individuals developing symptoms following initial treatment. The probability of experiencing recurrent symptoms while treated conservatively is based on the findings of the two RCTs reported in the systematic review of clinical effectiveness (Chapter 4). The estimate was derived from the number of patients in conservative management who switched to surgery or required further medical treatment during the 5-year follow-up period. 46 Forty-five out of 102 symptomatic people received further treatment, mainly surgery, over a 5-year period after randomisation to conservative management. The 5-year probability of presenting for surgery was converted into a fixed instantaneous rate and was reconverted into an annual probability using the following formula:133

where p is the probability, R is the rate, t is the time period of interest and ln is natural log.

The rate based on above data is therefore:

The 1-year probability of having symptoms is therefore:

The model was specified so that the proportion of patients remaining symptom free under conservative management increased over time (as those continuing to experience symptoms presented for surgery). As such, it was assumed that all those individuals remaining in conservative management at 5 years were symptom free.

Risk of death attached to surgery

This information was based on estimates of risk modelling for the probability of death after cholecystectomy using simulations for people that received elective surgery. 134 The estimate used in the base-case model was 0.00073 (Table 7).

All-cause mortality rates in the UK

The probability that people with gallstone disease entering the model may die is based on the annual rates of all-cause age-specific mortality for both men and women (Table 8). 118

Resource utilisation and cost estimation

Costs are presented in pounds sterling (£) in 2011–12 prices. Details of the sources used are summarised in Table 9. Data included the direct health service costs associated with each treatment strategy. The following resource use data were required to estimate costs incurred by the NHS: staff time, consumables, overheads and all the resources associated with any subsequent patient follow-up and management. Surgical procedures were mapped to their appropriate health-care resource group (HRG) and costed using NHS Reference Costs84 or the payment by results national tariff. The unit cost data for other resource utilisation was extracted from the literature or obtained from other relevant sources [e.g. Personal Social Service Research Unit].

The cost of the surgery was based on the cost of laparoscopic cholecystectomy with the length of stay of 1 day without complications. Laparoscopic cholecystectomy was chosen as it is the most frequently performed surgical procedure for people presenting with cholecystitis or biliary pain. 84 There are two main dimensions of outcome after laparoscopic cholecystectomy: surgical complications or no complications. The cost of peri-operative complications was assumed to be the same for both surgical procedures, open and laparoscopic. Additional costs were considered for those people whose complications required further surgical interventions such as the treatment of bile duct injury. These additional costs were based on the average cost of all services relating to hepatobiliary procedures with and without complications. It was also assumed that these people had four follow-up visits to the hepatobiliary clinic every year. Costs were also included for people who experienced pain after cholecystectomy (post-cholecystectomy syndrome). People who had no surgical complications could either be completely cured or alternatively could have suffered from persistent symptoms (post-cholecystectomy syndrome). Based on clinical advice the cost of unresolved symptoms was assumed to involve one general practitioner and one outpatient visit per year.

The resources needed to provide conservative management included the number of visits to different health care providers, in particular the general practitioner (GP) and hospital outpatient department. In the base-case analysis it was assumed that all people presenting with biliary pain or cholecystitis had one hepatobiliary consultant clinic visit after the initial presentation (all people in the total cohort had one initial presentation at clinic). People whose symptoms were not resolved and continued with conservative management were assumed to have two additional visits within the first year of presentation to secondary care (one to the GP and a follow-up visit to the hepatobiliary clinic) and two similar visits each year in the following cycles. The remainder of people whose symptoms were not resolved were assumed to have had laparoscopic surgery and followed the surgery care pathway. Pain relief medications were not considered in the base-case analysis as these costs are either incurred by the people themselves (rest of the UK) or are free (Scotland).

Estimation of total cost of strategies

The total costs of the treatment strategies were determined by using recursive cost in the Markov model. As the model tree expands from the left to the right, the ‘cost’ variable is modified by adding new cost variables. In this way the value of ‘cost’ at each terminal node is unique to the path from the root node to that terminal node. Discounted costs are considered in the model to estimate the cost of each strategy by using the following formula:

Quality-of-life measures

Quality-of-life weights associated with the different outcomes of treatment and follow-up of people with symptomatic gallstones were identified in order to extend the economic evaluation into a cost–utility analysis. A focused search of the Cost-effectiveness Analysis Registry (Appendix 1) was performed to identify QoL data relevant to a UK setting. The search indicated that there were 15 studies reporting utility weights. None of these studies compared conservative management with cholecystectomy. There were six papers that compared outcomes of different interventions such as mini laparotomy, lithotripsy, open cholecystectomy with laparoscopic cholecystectomy116,117,136–139 and one paper compared the different methods of valuing health-related QoL techniques, i.e. visual analogue scale and the standard gamble in measuring people preferences for outcomes of gallstone disease. 140

Two of the studies117,139 that were conducted alongside RCTs used the European Quality of Life-5 Dimensions (EQ-5D). Although Nilsson and colleagues139 collected data at five different time points (pre-operatively, post-operatively, 1 week, 1 month and 1 year), they did not report preference-based weights or estimate a QALY. Macafee et al. 117 compared early with delayed cholecystectomy and reported utility scores of 0.85 (standard deviation 0.26) for the early group and 0.93 (standard deviation 0.13) for the conventional (delayed) group (QALYs estimated using the EQ-5D questionnaire were calculated 30–35 days after laparoscopic surgery). The final study138 estimated QALYs for different health states relating to two treatment options using a time trade-off method. The utility score for laparoscopic cholecystectomy was found to be similar to those of the Macafee et al. 117 study, 0.90 (95% CI 0.87 to 0.93). In the model, utility values were assigned to the types of potential different morbidities that could be associated with the procedures or health state represented in the model. Table 10 provides a summary of the estimates used in the model. Estimates of the utility for people who had cholecystectomy were based on the results of a RCT. 117 The utility estimate for people whose symptoms were resolved using laparoscopic cholecystectomy or conservative management was 0.93. The rest of the estimates were based on a study that used standard gamble to quantify how much a given type of morbidity would detract from their QoL. 140 The estimation of QALY values took into account the time people stayed in the state. For example, individuals who had further surgery were in a reduced state of health and were assigned a utility value of 0.64 for a month and 0.80 for the rest of the year. By applying utility weights to the modelled states and temporary events within states in the Markov model, QALYs gained over the 5-year follow-up period were tracked.

Assessment of cost-effectiveness

The base-case analysis was based on the costs and outcomes of a cohort of 51-year-old symptomatic women (the mean age of people receiving cholecystectomy in England and Wales). The outcome is the incremental cost per QALY. Data on the outcome are presented in two ways. First, the mean costs, QALYs and the incremental cost-effectiveness ratio (ICER), where appropriate, are calculated. These are ratios of the differences in costs and interventions divided by the differences in clinical effectiveness between the different strategies. These data reflect the rate of return in QALYs to the quantity of resources used (measured in monetary values). The second way in which the cost-effectiveness of the alternative care pathways is presented is in terms of cost-effectiveness acceptability curves (CEACs). CEACs which are derived from probabilistic sensitivity analysis are used to illustrate the uncertainty caused by the statistical variability in the parameters used in the model. CEACs illustrate the likelihood that a strategy is cost-effective at various threshold values for society’s willingness to pay for an additional QALY. The value society is willing to pay for a QALY is unclear, but in the UK NICE tends to recommend that an intervention is cost-effective when the incremental cost per QALY is less than £20,000–30,000. 119

Probabilistic and deterministic sensitivity analyses

Probabilistic and deterministic sensitivity analyses were applied to the model in order to assess the robustness of the results to realistic variations in the model parameters. The point estimates applied to costs, outcomes and probabilities of events in the model are all estimates of true population values. As such, distributions are assigned to these parameters to characterise the surrounding uncertainty due to sampling variation. Monte Carlo simulation was used to repeatedly sample from the distributions assigned to the parameters (Tables 7, 9 and 10) and reanalyse the model a large number of times.

Probabilistic sensitivity analyses of the base case were conducted by assuming a beta distribution for the probability of successful treatment outcome in the conservative management strategy (the 5-year probability of presenting for surgery being the complement of this). All cost parameters were assigned a gamma distribution as this distribution fits the skewed nature found in resource use and cost data. For health-care resource group inpatient unit cost data, the interquartile range was used to derive the standard error. For the normal distribution this is defined by the interval ± 0.67 standard errors either side of the mean. 133 An estimate of the cost standard error was derived by generating the difference between the interquartile ranges and dividing it by 2 then multiplying by 0.6, where 0.67 is the critical value which excludes 50% of the standard normal distribution. There was no distribution attached to all-cause mortality rate as the estimate was derived from the general population. The utility parameters were assigned a beta distribution to take into account the constraints in terms of their construction, which is infinity at the lower end (worst possible health) and 1.0 at the upper end (perfect health).

Sensitivity analyses varied the assumptions and/or parameter point estimates applied in the base-case analysis. One-way sensitivity analyses (varying the probabilities, costs and QALY weights one at a time) were performed.

Cost of conservative management

There are no reliable data on the number of visits that people with symptoms of gallstone disease make to primary care providers before they are referred to a consultant, although it has been suggested that they may have multiple visits to their GP during the time they are awaiting treatment. The base-case analysis assumed that people with symptomatic gallstones had one primary care and one consultant visit. The number of visits was doubled to estimate the impact on cost-effectiveness. Another sensitivity analysis explored the inclusion of inpatient costs in the conservative management arm. According to the commissioner’s guide to cholecystectomy, it is not uncommon for a person to be readmitted as an emergency with a flare of symptoms three times during a prolonged period waiting for surgery. 90 A Canadian study on the risk of emergency while waiting for elective cholecystectomy reported that 6.7% of participants had emergency admissions. 141 Therefore, we estimated the cost of one, two and three additional inpatient stays for people who had persistent pain attacks while receiving conservative management. Further analysis was also performed assuming that all symptomatic people receiving conservative management had 1–3 non-elective episodes of short inpatient stay.

Utilities

As mentioned earlier, the data on utilities were based on published data and assumptions were made about the utility value of people who had received medical management. Although the RCTs included in the review of clinical effectiveness collected information on the number of pain attacks (within a pre-specified range) over a 5-year follow-up period, no data were provided on the duration of the pain attacks or on the number of pain attacks in each randomised group. Bass and colleagues140 estimated people preference values for selected outcomes by described frequency of pain or duration of outcome and their values ranged from 0.62 for having biliary pain once a day for a lifetime to 0.85 for having biliary pain once a week for 1 month. The CIs for these values spanned between 0.26 and 1.0 (full health).

Therefore, the utility of people that remained symptomatic in the conservative management arm was increased and decreased by 0.05 increments to explore the impact of this change on the results. Standard gamble utility weights for those who had open cholecystectomy with and without complications such as post-operative wound infections ranged from 0.57 to 0.83 for those with post-operative bile duct stone. Analysis was also performed to estimate the impact of increasing the utility of people who had continuing surgical complications from 0.64 to 0.80 (closer to the utility value of those patients who had open cholecystectomy 0.78).

Annual discount rate

The NICE guidelines recommend an annual discount rate of 3.5% for costs and outcomes and was applied in the base-case analysis. 119 A range of between 0% and 6% was considered in the sensitivity analyses.

Increasing follow-up to 10 years

Evidence from the long-term follow-up studies46,113 suggests that the number of people seeking further treatment reduces over time, so the base-case analysis considered the 5-year follow-up. A sensitivity analysis was performed to explore how the small variation in modelled health outcomes and ongoing management costs (at 5 years) influences cost-effectiveness over a longer time horizon (10 years).

Value of information analysis

The decisions related to the adoption or rejection of technologies should be based on expected cost-effectiveness given the current existing information. However, the decisions should not be based on little, or poor-quality evidence, as long as the decision to conduct further research to support adoption (or rejection) is made simultaneously. 132 Decisions to adopt or reject a technology based on existing information will be uncertain as there will always be a chance that the wrong decision could be made. Wrong decisions lead to opportunity costs due to forgone health benefits and resources. The expected cost of uncertainty is determined jointly by the probability that a decision based on existing information will be wrong and the consequences of a wrong decision. The estimates of the probability of error and opportunity costs of error can be used to calculate the expected cost of uncertainty or the expected loss surrounding the decisions. The expected cost of uncertainty can be interpreted as the expected value of perfect information (EVPI) and perfect information can be used to eliminate the possibility of making the wrong decision. We performed a value of information analysis to estimate the distribution of expected costs and expected QALYS of treatment of biliary pain/cholecystitis dependent on the uncertainty in the input parameters. The net monetary benefit (NMB) for the two strategies from each Monte Carlo iteration was computed to represent the possible values of NMB for all the possible realisations of the uncertain parameters. The maximum NMB among all the strategies was then calculated to determine the total EVPI per subject. The expected value of a decision undertaken with perfect information was derived by averaging the maximum net benefit with perfect information and the maximum net benefit with the current information. We then estimated the expected value for the entire population that could potentially benefit from more research (population EVPI). To calculate the population EVPI, we assumed the effective lifetime of the treatment to be 10 years. Benefits to future people were discounted at 3.5%. The annual patient population that could benefit from the results of a future study was based on the number of people undergoing a gallbladder excision per annum in England (Hospital Episode Statistics data84), multiplied by the time period of 10 years.

Deterministic and probabilistic results

The cost-effectiveness analysis aggregates the average 5-year costs for the treatments and time spent in each of the various health states of the model. Table 11 shows the results for the base-case analysis for a hypothetical cohort of 51-year-old women with symptomatic gallstones. The table reports the performance of surgery compared with conservative management, which is less costly and less effective. On average, conservative management care pathway costs less but has lower benefits (QALYs) than the surgery pathway. Analysis performed using a cohort of 51-year-old men had lower QALYs (4.108 for conservative management and 4.21 for cholecystectomy) but the ICER was similar to that of the base-case analysis (ICER £12,178).

Probabilistic results

To account for the uncertainty surrounding ICER (for surgery vs. conservative management) resulting from the joint uncertainty surrounding all the model input parameters, probabilistic sensitivity analysis using Monte Carlo simulations was performed. The results of the probabilistic analysis are reported in the form of CEACs (Figure 28). The conservative management arm has a higher probability of being considered cost-effective for thresholds below £20,000 and the surgery arm has a higher probability of being considered cost-effective as the threshold increases beyond £20,000.

FIGURE 28.

Cost-effectiveness acceptability curve determined by society’s willingness to pay for an additional QALY. CM, conservative management.

Sensitivity analyses

Cost of conservative management

The ICER value reduced as the number of inpatient stays (under conservative management) increased (Table 14). When all the people that had symptoms in the conservative management strategy had an inpatient stay there was more than 50% chance of the surgery arm being considered to be cost-effective for the thresholds above £5000.

Changes in the number of outpatient and GP visits had very little impact on the results.

Utilities

The ICER results were sensitive to the utility values applied (Table 15). When we assumed that the utility decrement for people who remain symptomatic under conservative management was greater than in the base case, the ICER reduced in favour of surgery (this is as expected, as there is more potential for health improvement if one’s utility value is lower). However, for values above the base case, the results favour conservative management as QALY gains with surgery decrease.

The analyses which we performed to estimate the impact of increasing the utility of people who had continuing surgical complications had a low impact on the overall results. On average, both surgery costs and benefits were higher, and led to a slightly lower ICER than that of the base-case analysis (£14,761).

Annual discount rate for costs and benefits

Varying the discount rate for costs did not change the results substantially (Table 16). However, the ICERs appeared to vary more when the utility discount rate was changed. As expected, the ICER increased as the rates increased.

Increasing follow-up to 10 years

The results moved in the same directions as those of the base-case analysis as the probability of getting surgery in the conservative management strategy only increased by 0.01. The evidence from the systematic review of clinical effectiveness indicated that there was only one additional person that received surgery from the conservative management arm in the 14-year follow-up period bringing the total to 46 out of 102 (45%). There was quite a substantial drop in QALY gain over the longer follow-up period (Table 17). This was probably because, after 5 years, people on conservative management were no longer symptomatic.

Value of information

The probabilistic analysis indicated that, for a threshold of £30,000, surgery had a 57% chance of being considered cost-effective. The expected saving from perfect information per person was £585.91 and the expected gain in effectiveness was 0.054 per person. The per person EVPI was £2209.65 and the resulting population EVPI was £1,206,730,653.

Clinical effectiveness

The two included studies46,113 we identified in the current literature were good-quality RCTs, which randomised a total of 201 participants with symptomatic gallstone disease to either cholecystectomy or observation. Both trials reported relevant outcomes, which were combined in mathematical syntheses when appropriate. It is noteworthy that both trials were conducted in Norway and not in the UK. We do not know how well the health-care system in Norway mirrors the UK NHS.

Gallstone disease associated events

On the whole, significantly more people randomised to observation (14%) than to surgery (2%) experienced gallstone-related complications. New events that occurred in the observation group were acute cholecystitis, common bile duct stones and acute pancreatitis. In particular, further complications occurred more frequently (27%) in people with an initial diagnosis of acute cholecystitis46 than in those (1%) with uncomplicated symptomatic gallstones (biliary pain only). 113 Few people suffered from CBD stones during the 14-year follow-up with no significant difference between intervention groups (4% in the observation group vs. 1% in the surgery group). In particular, CBD stones occurred in only one person with uncomplicated symptomatic gallstones (biliary pain only) rather than acute cholecystitis during the 14-year follow-up period. Similarly, the proportion of people suffering from acute pancreatitis was the same in both intervention groups (1% observation group vs. 1% in the surgery group). These findings are in line with those of early natural history studies which report a cumulative rate of 2–6% for CBD stones or acute pancreatitis in participants observed for 5–10 years. 17,42,92 The overall risk of developing acute biliary complications in people with symptomatic gallstones has been estimated to be around 3% per annum during an observation period of 6 years. 143 In a recent population-based study,12 gallstone-related complications were observed in 7.5% and 22% of mild and severe symptomatic people, respectively. In short, in the included RCTs, the majority of people treated conservatively did not experience long-term complications (86%); CBD stones and episodes of acute pancreatitis were both rare.

Pain attacks

Overall, the proportion of people suffering from pain attacks was not significantly different between randomised groups (25% in the observation group vs. 12% in the surgery group). Pain attacks were observed more frequently in participants treated conservatively (26/102 people) than in those who underwent surgery (12/99 people) and in participants with uncomplicated symptomatic gallstones (biliary pain only – 23/69 people) rather than in those with an initial diagnosis of acute cholecystitis (3/33). The admissions to hospital as a result of pain were not significantly different between intervention groups. However, admissions were more frequently observed in people with uncomplicated symptomatic gallstones (biliary pain only – 12/69) rather than in those with acute cholecystitis (4/33). It is noteworthy that, when we considered people with gallstones according to the treatment they actually received and not according to randomisation, fewer episodes of pain were observed in people treated conservatively (12/68) than in those who were eventually operated on (26/133).

The review findings on pain attacks are broadly similar to those suggested in the clinical literature, i.e. cholecystectomy is not always successful in relieving symptomatic people from pain. Up to 40% of people may suffer from post-cholecystectomy symptoms. 56 Symptoms may persist or arise de novo after surgery. 63 Marked biliary pain has been described in 4–9% of people after cholecystectomy, while persistent abdominal pain or non-specific pain has been reported in about 13–37% of people. 57–62 Some investigators have also reported a persistent pain similar to that experienced pre-operatively in approximately 20% of people after cholecystectomy. 64,65

Surgery rate

The majority of people randomised to receive surgery (88%) and approximately half of the people randomised to observation (45%) eventually underwent cholecystectomy during the 14-year follow-up period. As few gallstone-related complications were experienced by symptomatic people randomised to observation, it is likely that pain was the reason why they underwent surgery. 114 Most of the surgical procedures were performed during the first 5 years and virtually no operations occurred after 5 years. These data are in accordance with previously published studies. McSherry and colleagues17 found that 44% of people with mild gallstone disease underwent surgery during a 5-year follow-up period. Similarly, Festi and colleagues12 reported that 17% and 41% of people with mild and severe symptoms, respectively, underwent surgery during a 10-year follow-up period, but the majority of the surgical procedures (68%) were performed during the first 4 years.

Not all participants in the observation group required cholecystectomy or experienced severe complications during the long-term follow-up. Interestingly, 55% of people randomised to observation did not require an operation during the 14-year follow-up period and 12% of people randomised to receive cholecystectomy did not undergo the scheduled operation, probably as a result of minimal, tolerable symptoms. These findings are in agreement with those of early natural history studies17,45 and with those of more recent observational and population-based studies. 12,56 Larsen and colleagues56 found that 45% of symptomatic people on watchful waiting were totally relieved of symptoms during a 1-year observation period. Similarly, Festi and colleagues12 observed that 58% of people with initially mild symptoms and 52% of those with more severe symptoms did not experience further episodes of pain during a follow-up period of 10 years.

Surgery associated events

The risk of surgery-related complications including abdominal infection, bile leakage, wound infection, dehiscence and bile duct injury was significantly higher in the cholecystectomy group. Nineteen per cent of participants originally randomised to surgery, compared with 7% participants randomised to observation (who were eventually operated on), experienced surgery-related complications. Complications of cholecystectomy have been well documented and the findings of our two RCTs46,113 are consistent with those of three recent Cochrane systematic reviews106,144,145 which indicate that the total number of people suffering complications after open, small-incision or laparoscopic cholecystectomy is high with no significant differences between the three surgical procedures. 106 In particular, the proportions of total complications after laparoscopic and small-incision cholecystectomy calculated from 13 clinical trials at low risk of bias (total 2337 participants) were 17% and 17.5%, respectively. 106

Mortality rate

There were no deaths caused by gallstone disease in the included trials. The all-cause mortality at 14 years was greater (but not significantly greater) among participants randomised to surgery. Even though a significant difference between intervention groups was not detected, it is noteworthy that the relative risk was 0.73 (95% CI 0.41 to 1.31) favouring the observation group. This finding suggests that people undergoing surgery had an increased probability of death compared with non-operated people. A clear explanation of this finding is difficult to determine. However, surgery in itself carries an intrinsic mortality risk, especially if there are pre-existing comorbidities.

Quality of life

We were not able to meta-analyse the scores of QoL scales. The trial investigators reported that, overall non-significant differences were observed between participants who underwent cholecystectomy and those treated conservatively, and no gallstone or surgery-related deaths occurred in the two randomised groups. Participants in both treatment groups experienced a decline over time in pain scores, and their QoL scores improved after randomisation. It is noteworthy that participants with more frequent and marked pain at randomisation were reported to have a higher probability of subsequent gallstone-related events. For the group with an initial diagnosis of acute cholecystitis, the trial investigators analysed the results according to the treatment participants actually received and unexpectedly did not find any difference between observed and operated participants. It is likely, however, that the sample size of the two groups (64 participants in total) was too small to detect statistically significant differences.

Early compared with delayed surgery

We identified only two RCTs in the current literature, which compared observation/conservative management with cholecystectomy. Several RCTs have been conducted on the effects and safety of early compared with delayed surgery. Although these studies do not focus specifically on people with uncomplicated symptomatic gallstone disease or with a first episode of biliary pain presenting to secondary care, and the delayed period is relatively short (a few weeks), it is worth considering the outcomes of people allocated to ‘delayed surgery’ in terms of complications and need for emergency surgery in light of our review results. In particular, two Cochrane systematic reviews have been recently published on the effects of early laparoscopic cholecystectomy compared with delayed cholecystectomy in people with either acute cholecystitis or biliary pain. 72,146

The Cochrane systematic review on early laparoscopic cholecystectomy (less than 7 days from symptom onset) compared with delayed cholecystectomy (more than 6 weeks after initial admission) in people with an initial diagnosis of acute cholecystitis146 included five trials with a total of 451 participants. No statistically significant differences between the two groups were detected for any of the outcomes considered, including complication rate (bile duct injury odds ratio 0.6; 95% CI 0.15 to 2.70) and conversion to open cholecystectomy (odds ratio 0.85; 95% CI 0.53 to 1.34). Length of follow-up was not reported. However, 18% of the people randomised to the delayed group had non-resolution or recurrence of symptoms before their planned operation and, therefore, required emergency laparoscopic cholecystectomy. The proportion of these emergency procedures converted to open cholecystectomy was 45%.

The same authors conducted a systematic review to evaluate the benefits and harms of early compared with delayed laparoscopic cholecystectomy in participants with biliary pain due to gallstone disease. 72 They identified only one trial with a total of 75 participants in the literature. 87 Morbidity, and in particular severe morbidity (e.g. acute pancreatitis, bile duct injury, gallbladder perforation), was significantly lower in the early group (0%, 0/35) than in the delayed group (15%, 6/40). Overall, during the delayed period (mean 4.2 months) 35% of people in the delayed group (14/40) required 18 hospital admissions for symptoms related to gallstone disease (hospital admissions for people in the early group were not reported). Hospital stay was significantly shorter in the early group than in the delayed group (weighted mean difference −1.25 days; 95% CI −2.05 to −0.45). The trial was at high risk of bias (unclear allocation concealment, lack of blinding, no sample size calculation and no intention-to-treat analysis) hence its results can only be interpreted tentatively.

Factors influencing the decision to undertake a cholecystectomy

From a clinical point of view, the review findings suggest that it would be crucial to identify which characteristics of gallstone disease could predict the evolution of symptoms. In particular, it would be very important to be able to single out people with uncomplicated symptomatic gallstones who could be treated conservatively. Unfortunately, at present no definitive factors have been identified. Even the presence of pain in the upper right abdominal quadrant may not consistently be related to gallbladder stones147 and, overall, the symptoms caused by gallstone disease are still not completely understood, rendering it difficult to identify people who could benefit from surgery. Furthermore, people vary considerably in the way they report their symptoms depending on their individual pain threshold as well as their preference for having or avoiding surgery. At present, in many cases, surgery is offered independently from the clinical symptoms of people presenting with gallstone disease and without considering the potential benign course of the disease partly due to the wide diffusion of laparoscopic cholecystectomy. According to the World Gastroenterology Organisation Practice Guideline148 on gallstone disease, when a group of nine surgeons were asked to evaluate 252 people who underwent cholecystectomy, they agreed that surgery was the appropriate option in only 52% of cases and failed to find an agreement on 44% of cases. There is now an indication that removal of the gallbladder may also potentially lead to a slightly increased risk of colon cancer in women in the long term. 149,150

Summary

In our assessment, approximately half of the people in the observation group were eventually operated on. Almost no surgical operations were performed after the first 5 years of follow-up. Participants who underwent cholecystectomy experienced more surgery-related complications and showed a slight, non-significant, increase in mortality rate (all-cause mortality) than those who were treated conservatively. Conversely, participants in the observation groups had more episodes of cholecystitis, but the incidence of other gallstone-related complications (e.g. common bile duct stones, acute pancreatitis) was low. Approximately half of the people in the observation group did not required surgery at long-term indicating that there is probably a subgroup of people with symptomatic gallstone disease who could be safely treated conservatively.

Summary of findings from the economic model

The economic model was developed using evidence synthesis from three main sources: a systematic review and meta-analysis of clinical effectiveness, advice from clinical experts and a prior health state valuation survey. The economic model compared two treatment strategies: conservative management and surgery. The model compared cumulative costs and QALYs for a cohort of 51-year-old women for a 5-year time horizon. This time was chosen as the available data indicated that the results did not markedly change if a longer-term horizon was adopted. On average, the surgery strategy cost £1236 more than the conservative management strategy but was, on average, more effective, and generated 0.094 additional QALYs. The incremental cost per QALY was £13,205. The incremental cost-effectiveness result indicated that the conservative management strategy had a 51% chance of being considered cost-effective at society’s willingness-to-pay threshold of £20,000. The probability of cost-effectiveness was not sensitive to changes in willingness to pay, when the threshold increased to £30,000, the conservative management strategy had a 46% chance of being considered cost-effective.

The result of the analysis using a male cohort aged 51 years was similar to that of the base-case analysis. On average, surgery cost more but was more clinically effective, even though the effectiveness for both arms was slightly lower (conservative management 4.108 and cholecystectomy 4.21) leading to an ICER of £12,718. This slight reduction in QALYs was the result of the higher mortality of the male cohort.

The results were sensitive to the probability of the number of people needing surgery in the conservative management strategy. The systematic review indicated that there was uncertainty around the estimate derived in the meta-analysis. On average, the cost of the conservative management strategy reduced to £694 when the probability was reduced to 25% leading to an ICER of £33,542 per QALY for the surgical strategy. The probability that the conservative management strategy would be considered cost-effective was between 56% and 61% (for willingness to pay for QALY values of £20,000–30,000, respectively). However, the ICER reduced to £4291 when the probability of surgery among those initially managed conservatively was increased to 75%. The probability that the surgery strategy would be considered cost-effective at the recommended threshold was between 51% and 53%.

There was no evidence to indicate that the risk of surgical intervention increased over time for patients that required surgery after conservative management. Therefore, these patients were assumed to follow a similar pathway to those who went straight into surgery. However, the model considered that patients who had surgery after conservative management had a lower utility prior to surgery – to take into account the fact that they remained symptomatic.

Evidence suggests that the number of complications from laparoscopic cholecystectomy has reduced over the years as the procedure has become widely used. The sensitivity analyses results around the probability of surgical complications were similar to those of the base-case analysis (£13,739 when probability of complications = 0.05, and £12,135 when probability of complications = 0.2), suggesting that the model is robust to the changes in the probability of surgical complications.

The results were more sensitive to changes in the costs of conservative management. As there are no data to support the average number of visits people may have to the GP and the hospital, various exploratory analyses were performed to assess the impact of additional inpatient stays for people in the conservative management strategy. As anticipated, as the number of inpatient stays (each year) for the group with persistent pain increased, the cost of the conservative management strategy increased, thereby reducing the cost difference between the two strategies. This led to a reduction of the ICER from £13,205 (no inpatient stay) to £1253 (three non-elective inpatient short stays).

The results were most sensitive to the utility value that was used for people who were managed conservatively. Based on a previous valuation survey, in the base-case analysis it was assumed that the utility was 0.71 among those with persistent symptoms and 0.80 for those with no persistent pain. As appropriate information on the number of pain attacks and their impact on QoL was not available from the trial data, these values could have been over- or underestimated, and were therefore scrutinised in sensitivity analysis. The ICERs varied from £4175 (when utility value equalled 0.60 for persistent symptoms), to conservative management strategy dominating surgery when the utility value for persistent symptoms was assumed to be 0.9. The probability of conservative management strategy being considered cost-effective for society’s willingness-to-pay thresholds of £20,000–30,000 ranged from 8% to 82%. With changes in the utility value of persistent symptoms from 0.60 to 0.90, the probability of the surgery strategy being considered cost-effective at willingness-to-pay threshold of £20,000 changed from 92% to 18%. The results were not sensitive to changes in the utility of complications associated with surgery. The results were not markedly sensitive to changes in the discount rates applied to both the costs and the benefits.

Databases

Trials

Conference proceedings from:

Association of Upper Gastrointestinal Surgeons of Great Britain and Ireland.

Association of Laparoscopic Surgeons of Great Britain and Ireland.

British Society of Gastroenterology.

Databases

Databases

Schmidt et al. 2011113

Schmidt M, Søndenaa K, Vetrhus M, Berhane T, Eide GE. A randomized controlled study of uncomplicated gallstone disease with a 14-year follow up showed that operation was the preferred treatment. Dig Surg 2011;28:270–6.

Schmidt et al. 201146

Schmidt M, Søndenaa K, Vetrhus M, Berhane T, Eide GE. Long-term follow-up of a randomized controlled trial of observation versus surgery for acute cholecystitis: Non-operative management is an option in some patients. Scand J Gastroenterol 2011;46:1251–6.

Ineligible comparator

Ahmad I. Cholecystectomy in acute cholecystitis. J Pak Med Assoc 1992;42:112–15.

Ajao OG, Al-Saigh AA, Malatani T, Ladipo JK. Timing of surgery in acute cholecystitis. East Afr Med J 1991;68:490–5.

Al-Khuwaitir S. Early versus delayed surgery for acute cholecystitis. J Pak Med Assoc 1984;34:210–13.

Banz VM, Gsponert T, Candinas D, Guller U. Population-based analysis of 4126 patients with acute cholecystitis: Defining the optimal time-point for laparoscopic cholecystectomy. Surg Endosc 2011;25:S2.

Cameron IC, Chadwick C, Phillips J, Johnson AG. Acute cholecystitis – room for improvement? Ann R Coll Surg Engl 2002;84:10–13.

Chang TC, Lin MT, Wu MH, Wang MY, Lee PH. Evaluation of early versus delayed laparoscopic cholecystectomy in the treatment of acute cholecystitis. Hepato-Gastroenterology 2009;56:26–8.

Eitan A, Toledano C, Rivlin E, Linn S, Barzilai A. Early vs. delayed cholecystectomy for acute cholecystitis. Harefuah 1991;120:319–68.

Fowkes FG, Gunn AA. The management of acute cholecystitis and its hospital cost. Br J Surg 1980;67:613–17.

Gomez-Ferrer F, Ortega J, Cortes J, Ruiz J, Arnal C. Early or delayed operations in acute cholecystitis. Rev Esp Enferm Apar Dig 1988;73:259–63.

Isoda N, Ido K, Kawamoto C, Suzuki T, Nagamine N, Ono K, et al. Laparoscopic cholecystectomy in gallstone patients with acute cholecystitis. J Gastroenterol 1999;34:372–5.

Jaeger G, Rothenbuhler JM, Famos M, Tondelli P. When should cholecystectomy in acute cholecystitis be planned? Schweiz Med Wochenschr 1983;113:552–4.

Jarvinen HJ, Hastbacka J. Early cholecystectomy for acute cholecystitis: a prospective randomized study. Ann Surg 1980;191:501–5.

Johansson M, Thune A, Lundell L. A prospective randomised trial comparing early versus delayed laparoscopic cholecystectomy in the treatment of acute cholecystitis. Gastroenterology 2002;123:24.

Johansson M, Thune A, Blomqvist A, Nelvin L, Lundell L. Management of acute cholecystitis in the laparoscopic era: results of a prospective, randomized clinical trial. J Gastrointest Surg 2003;7:642–5.

Johansson M, Thune A, Blomqvist A, Nelvin L, Lundell L. Impact of choice of therapeutic strategy for acute cholecystitis on patient’s health-related quality of life. Results of a randomized, controlled clinical trial. Dig Surg 2004;21:359–62.

Kontopodis N, Spiridakis K, Panagiotakis G, Grigoraki M, Kokkinakis T, Rokadakis L. Acute cholecystitis: conservative treatment and delayed cholecystectomy. Chirurgia 2011;24:121–4.

Kozelj M, Flis V, Potrc S, Horvat M, Mursak M. Comparison between delayed and early cholecystectomy in acute cholecystitis. J Hepatol 1996;25:209.

Lai PB, Kwong KH, Leung KL, Kwok SP, Chan AC, Chung SC, et al. Randomized trial of early versus delayed laparoscopic cholecystectomy for acute cholecystitis. Br J Surg 1998;85:764–7.

Lawrentschuk N, Hewitt PM, Pritchard MG. Elective laparoscopic cholecystectomy: implications of prolonged waiting times for surgery. ANZ J Surg 2003;73:890–3.

Lin A, Stiven P, Bagshaw P, Connor S. Cholecystectomy following acute presentation to a major New Zealand metropolitan hospital: change to the timing of surgery is needed. N Z Med J 2006;119:U2104.

Mare LD, Saadi A, Roulin D, Demartines N, Halkic N. Delayed versus early laparoscopic cholecystectomy for acute cholecystitis: A prospective randomized study. HPB 2012;14:130.

Maroske D, Röher HD. Early but not delayed cholecystectomy in acute cholecystitis. Dig Surg 1984;1:98.

Navarra S, Melita P, Calbo L, Barbuscia M, Palmeri R, Lorenzini C. Acute cholecystitis. Immediate or postponed treatment? Minerva Chir 1983;38:1391–6.

Norby S, Herlin P, Holmin T. Early or delayed cholecystectomy in acute cholecystitis? A clinical trial. Br J Surg 1983;70:163–5.

Papi C, Catarci M, D’Ambrosio L, Gili L, Koch M, Grassi GB, et al. Timing of cholecystectomy for acute calculous cholecystitis: a meta-analysis. Am J Gastroenterol 2004;99:147–55.

Paran H, Zissin R, Rosenberg E, Griton I, Kots E, Gutman M. Prospective evaluation of patients with acute cholecystitis treated with percutaneous cholecystostomy and interval laparoscopic cholecystectomy. Int J Surg 2006;4:101–5.

Sanchez Bueno F, Parrilla P, Ayllon JG, Lujan JA, Robles R, Ramirez P, et al. Timing of operation in acute cholecystitis: Prospective study of 375 cases divided into three series – early, delayed and late surgery. Dig Surg 1993;10:72–7.

Schaefer D, Barth H, Thon K, Jostarndt L, Maroske D. Early or delayed operation in patients with acute cholecystitis. Results of a prospective randomized controlled clinical trial. Chir Forum Exp Klin Forsch 1980;149–53.

Serralta AS, Bueno JL, Planells MR, Rodero DR. Prospective evaluation of emergency versus delayed laparoscopic cholecystectomy for early cholecystitis. Surg Laparosc Endosc Percutan Tech 2003;13:71–5.

Siddiqui T, MacDonald A, Chong PS, Jenkins JT. Early versus delayed laparoscopic cholecystectomy for acute cholecystitis: a meta-analysis of randomized clinical trials. Am J Surg 2008;195:40–7.

van der Linden W, Edlund G. Early versus delayed cholecystectomy: the effect of a change in management. Br J Surg 1981;68:753–7.

Williamson WA, Simpson CJ, Mclatchie G, Gillespie C, McArdle CS. Prospective randomized trial of early vs. delayed surgery for acute biliary tract disease. Gut 1984;25:A1140–1.

Zargar-Shoshtari K, Short H, Poole GH, Hill AG. Acute laparoscopic cholecystectomy: preferred treatment for acute biliary disease. ANZ J Surg 2008;78:771–4.

Ineligible study design

Al-Saleh MS, Almuhiza I, Alharthy M. Early cholecystectomy for acute cholecystitis. Saudi Med J 1995;16:227–30.

Contini S, Corradi D, Busi N, Alessandri L, Pezzarossa A, Scarpignato C. Can gangrenous cholecystitis be prevented?: a plea against a ‘wait and see’ attitude. J Clin Gastroenterol 2004;38:710–16.

D’Ambrosio R, Leone L, Peppas C, Grasso A, Di M, D’Ambrosio E, et al. Acute cholecystitis: timing of surgery. Minerva Chir 1995;50:39–46.

Festi D, Reggiani ML, Attili AF, Loria P, Pazzi P, Scaioli E, et al. Natural history of gallstone disease: Expectant management or active treatment? Results from a population-based cohort study. J Gastroenterol Hepatol 2010;25:719–24.

Gambiez L. How should biliary lithiasis be managed? Gastroenterol Clin Biol 2001;25:S128–39.

Ghani AA, Jan WA, Ul H. Acute cholecystitis: Immediate versus interval cholecystectomy. J Postgrad Med Inst 2005;19:192–5.

Goldfinger SE. By the way, doctor... Last month I had a night of severe abdominal pain. My doctor ordered an ultrasound test, and it showed gallstones. She is recommending surgery to remove my gallbladder, but I’d rather not have an operation because of this one episode. Is it dangerous to hold off on surgery? Is there any diet to follow or medicine I can take to prevent another attack? I am 53 years old and healthy. Harv Health Lett 1999;24:3.

Jorgensen T. Treatment of patients with uncomplicated gallbladder stones? Scand J Gastroenterol 2002;37:745.

Kubo G, Mann H, Nicko HJ. Treatment of acute cholecystitis by comparing conservative therapy, interval operation and early operation. Zentralbl Chir 1985;110:1054–65.

Malik A, I, Tou S, Nelson RL. Early laparoscopy versus observation for patients with acute abdominal pain. Cochrane Database Syst Rev 2009;1:CD008035.

Petakovic G, Korica M, Gavrilovic S. Acute cholecystitis – early or delayed cholecystectomy. Med Pregl 2002;55:135–9.

Ranalli M, Testi W, Genovese A, Bing C, Tumbiolo S, Andolfi E, et al. Early vs conservative treatment of acute cholecystitis. Personal experience and review of the literature. Minerva Chir 2004;59:547–53.

Sackmann M. Conservative treatment of gallstones: pro. Dtsch Med Wochenschr 2000;125:1051–2.

Ineligible population

Arthur JD, Edwards PR, Chagla LS. Management of gallstone disease in the elderly. Ann R Coll Surg Engl 2003;85:91–6.

Boerma D, Rauws EA, Keulemans YC, Janssen IM, Bolwerk CJ, Timmer R, et al. Wait-and-see policy or laparoscopic cholecystectomy after endoscopic sphincterotomy for bile-duct stones: a randomised trial. Lancet 2002;360:761–5.

Boerma D, Rauws EAJ, Keulemans YCA, Janssen IMC, Bolwerk CJM, Timmer R, et al. More recurrent complaints with a wait-and-see approach than with laparoscopic cholecystectomy after endoscopic sphincterotomy for common bile duct stones; randomised clinical trial. Ned Tijdschr Geneeskd 2002;146:1786–92.

Calero Garcia P, Ruiz Tovar J, Sanjuanbenito Dehesa A, Calero Amaro A, Diez Tabernilla M, Latorre Fragua R, et al. Acute cholecystitis: is it still justified to delay surgery? Cir Esp 2010;88:92–6.

Chiedozi LC, Al H, Salem MM, Al M, Okpere EE. Management of symptomatic cholelithiasis in pregnancy. Ann Saudi Med 2001;21:38–41.

Daradkeh S, Sumrein I, Daoud F, Zaidin K, Abu-Khalaf M. Management of gallbladder stones during pregnancy: conservative treatment or laparoscopic cholecystectomy? Hepato-Gastroenterology 1999;46:3074–6.

Elamin Ali M, Al-Shehri MY, Abu-Eshy S, Cheema MA, Mustafa Z, Sadek A. Is surgical intervention in acute cholecystitis in pregnancy justified? J Obstet Gynaecol 1997;17:435–8.

Gonzalez Gonzalez JJ, Sanz Alvarez L, Grana Lopez JL, Bermejo Abajo G, Navarrete Guijosa F, Martinez Rodriguez E. Gallstone disease in patients over the age of 80. Surgery or long-term medical treatment? Rev Esp Enferm Dig 1997;89:196–205.

Irvin TT, Arnstein PM. Management of symptomatic gallstones in the elderly. Br J Surg 1988;75:1163–5.

Keulemans YC, Rauws EA, Huibregtse K, Gouma DJ. Current management of the gallbladder after endoscopic sphincterotomy for common bile duct stones. Gastrointest Endosc 1997;46:514–19.

Kuy S, Roman SA, Desai R, Sosa JA. Outcomes following cholecystectomy in pregnant and nonpregnant women. Surgery 2009;146:358–66.

Lau J, Leow CK, Fung TMK, Suen BY, Yu LM, Lai PBS, et al. Five year follow-up of a randomized comparison of laparoscopic cholecystectomy or gallbladder in situ after endoscopic sphincterotomy and clearance of bile duct stones in patients older than 60. Gastroenterology 2005;128(Suppl. 2):A79.

Lu EJ, Curet MJ, El-Sayed YY, Kirkwood KS. Medical versus surgical management of biliary tract disease in pregnancy. Am J Surg 2004;188:755–9.

Plaisier PW, Nijs HG, van der Hul RL, den Toom R, Terpstra OT. 3-year results of ‘gallstone outpatient clinic’ in an academic hospital. Ned Tijdschr Geneeskd 1992;136:978–82.

Sanchez-Beorlegui J, Lamata-Hernandez F, Lagunas-Lostao E, Monsalve-Laguna E, Peñalva-Segura P. Choice of therapeutic approach for acute cholecystitis in the elderly. Rev Gastroenterol Mex 2010;75:149–57.

Swisher SG, Schmit PJ, Hunt KK, Hiyama DT, Bennion RS, Swisher EM, et al. Biliary disease during pregnancy. Am J Surg 1994;168:576–81.

Yamashita Y, Takada T, Hirata K. A survey of the timing and approach to the surgical management of patients with acute cholecystitis in Japanese hospitals. J Hepatobiliary Pancreat 2006;13:409–15.

No relevant outcomes

Søndenaa K, Nesvik I, Solhaug JH, Soreide O. Randomization to surgery or observation in patients with symptomatic gallbladder stone disease. The problem of evidence-based medicine in clinical practice. Scand J Gastroenterol 1997;32:611–16.

Unavailable

Borodach VA, Borodach AV. Surgical treatment of cholelithiasis in elderly and aged patients. Khirurgiia 2002;11:38–41.

Dimov R, Strangev G, Kandilarov K, Marinov V, Argirov D, Velkov B, et al. Prospective trial of early and delayed laparoscopic cholecistectomy in patients presented with acute cholecystitis. Bulg Med 2005;13:6–9.

Fisher DA. Watchful waiting after endoscopic removal of common bile duct stones: cheaper and better? Am J Gastroenterol 2006;101:753–4.

Grinev MV. Time of the operations in acute cholecystitis. Vestn Khir Im I I Grek 1988;140:22–6.

Hernandez Estrada AI, Aquirre Osete X, Pedraza Gonzalez LA. Laparoscopic cholecystectomy in pregnancy. Five years experience in the Spanish Hospital of Mexico and literature review. Ginecol Obstet Mex 2011;79:200–5.

Khan SSA. Early versus delayed cholecystectomy for acute cholecystitis, a prospective randomized study. Pak J Gastroenterol 2002;16:30–4.

Lopez EV, Florez JB, Fernandez CH, Valens PE, Caperochipi JA. Acute cholecystitis 2. Comparative study of the surgical treatment – early, deferred, or elective. Cir Esp 1982;36:244–51.