Outcomes Following an Index Emergency Admission With Cholecystitis

Supplemental Digital Content is available in the text Objective: The objective of this study was to evaluate the differences between patients who undergo cholecystectomy following index admission for cholecystitis, and those who are managed nonoperatively. Summary Background Data: Index emergency cholecystectomy following acute cholecystitis is widely recommended by national guidelines, but its effect on clinical outcomes remains uncertain. Methods: Data collected routinely from the Hospital Episode Statistics database (all admissions to National Health Service organizations in England and Wales) were extracted between April 1, 2002 and March 31, 2015. Analyses were limited to patients aged over 18 years with a primary diagnosis of cholecystitis. Exclusions included records with missing or invalid datasets, patients who had previously undergone a cholecystectomy, patients who had died without a cholecystectomy, and those undergoing cholecystectomy for malignancy, pancreatitis, or choledocholithiasis. Patients were grouped as either “no cholecystectomy” where they had never undergone a cholecystectomy following discharge, or “cholecystectomy.” The latter group was then subdivided as “emergency cholecystectomy” when cholecystectomy was performed during their index emergency admission, or “interval cholecystectomy” when a cholecystectomy was performed within 12 months following a subsequent (emergency or elective) admission. Propensity Score Matching was used to match emergency and interval cholecystectomy groups. Main outcome measures included 1) One-year total length of hospital stay due to biliary causes following an index emergency admission with cholecystitis. 2) One-year mortality; defined as death occurring within 1 year following the index emergency admission with acute cholecystitis. Results: Of the 99,139 patients admitted as an emergency with acute cholecystitis, 51.1% (47,626) did not undergo a cholecystectomy within 1 year of index admission. These patients were older, with more comorbidities (Charlson Comorbidity Score ≥ 5 in 23.5% vs. 8.1%, P < 0.001) when compared to patients who did have a cholecystectomy. While all-cause 1-year mortality was higher in the nonoperated versus the operated group (12.2% vs. 2.0%, P < 0.001), gallbladder-related deaths were significantly lower than all other causes of death in the non-operated group (3.3% vs. 8.9%, P < 0.001). Following matching, 1-year total hospital admission time was significantly higher following emergency compared with interval cholecystectomy (17.7 d vs. 13 d, P < 0.001). Conclusions: Over 50% of patients in England did not undergo cholecystectomy following index admission for acute cholecystitis. Mortality was higher in the nonoperated group, which was mostly due to non-gallbladder pathologies but total hospital admission time for biliary causes was lower over 12 months. Increasing the numbers of emergency cholecystectomy may risk over-treating patients with acute cholecystitis and increasing their time spent admitted to hospital.


Structured Abstract
Objective The objective of this study was to evaluate the differences between patients who undergo cholecystectomy following index admission for cholecystitis, and those that are managed non-operatively.

Summary Background Data
Index emergency cholecystectomy following acute cholecystitis is widely recommended by national guidelines, but its effect on clinical outcomes remains uncertain.

Introduction
The prevalence of gallstones in the global adult population is between 10-15%. 1 While 80% of people do not suffer pain or complications, the most common complication is acute cholecystitis. 2 Laparoscopic cholecystectomy is generally recommended to prevent recurrent episodes of cholecystitis and further complications after the initial admission. 1 Clinical trials have concentrated on the timing of performing cholecystectomy in patients with acute cholecystitis and operative outcomes. 1, 3 Level one evidence advocates index emergency cholecystectomy for the majority of patients presenting with acute cholecystitis when compared with interval cholecystectomy. [3][4][5][6][7] Due to these perceived benefits, financial incentives are being offered to increase the numbers of index emergency operations performed across many healthcare systems, including the National Health Service (NHS) in the United Kingdom (UK). [8][9][10][11][12] Despite this, data from Europe, Asia and North America show that the majority of patients with acute cholecystitis still undergo interval cholecystectomy. [13][14][15][16] An area that is less well studied is the natural history of acute cholecystitis. Early studies showed that recurrent attacks of pain and complications may actually diminish in up to half of initially symptomatic patients. 17,18 One concern is that patients may be over-treated if policies continue to promote index emergency cholecystectomy. Furthermore, while a non-operative approach would clearly avoid surgery-related complications, it is unclear if delaying cholecystectomy has a detrimental effect on patient outcomes and increases healthcare costs.
The objective of this current study was to use the advantages of a large population database to examine the medium-term sequelae following a common entry point of patients admitted with P o s t -P r i n t 4 acute cholecystitis. The primary aim was to identify the patient and hospital variables that determine which patients do not undergo a cholecystectomy following index emergency admission for acute cholecystitis, and their 1-year mortality. The secondary aim was to understand if interval cholecystectomy on a subsequent admission had a detrimental effect on patients by comparing 1-year total hospital admission time and mortality with those patients treated with emergency cholecystectomy during index admission. P o s t -P r i n t

Data Sources
Data were extracted from the Hospital Episode Statistics (HES, http://www.hscic.gov.uk/hes) database, which contains patient demographic and clinical information on every admission to an NHS hospital in England. HES is linked to the Office for National Statistics (ONS), which collects data on all registered deaths in the UK, to obtain mortality information. Data were analyzed in line with a data sharing agreement with NHS Digital for HES data.

Variables
To allow comparable risk adjustment, patient and hospital characteristics were defined as explanatory variables. These included patient age, gender, ethnicity, quintiles of Index of Multiple Deprivation (IMD) score, 19 Charlson comorbidity score, 20

Statistical Analysis
Results presented here are as described in accordance with the STROBE guidelines for reporting observational studies. 22 Data were analyzed using Stata ® version 14 (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX, USA). Differences between groups were evaluated using chi-squared tests for categorical variables and t-tests for continuous variables.
Propensity score matching (PSM) was used to select patients with similar characteristics who underwent interval cholecystectomy with those who underwent emergency cholecystectomy.
Patients were matched on a one-to-one basis on multiple confounders (age, gender, ethnicity, IMD score, comorbidities, and trust volume). Greedy matching was performed using the nearest neighbor technique within a caliper distance of 0.2 standard deviations, without replacement. 23 Although PSM reduced the total population data available for analysis, this technique provides a more accurate estimate of treatment effect by reducing selection bias in these otherwise heterogeneous groups. 23 Logistic regression models were used to compare 1-year mortality between the two matched groups. Statistical significance was measured at p<0.05, with results expressed as adjusted odds ratios (ORs) or incidence rate ratios (IRR) with 95 per cent confidence intervals (CI). Negative binominal regression was used to compare total length of stay between matched groups. One-year survival following emergency admission for cholecystitis was assessed using the Kaplan-Meier method. The cholecystectomy and no cholecystectomy groups were compared using the log-rank test.

Cholecystectomy vs. non-operative treatment for acute cholecystitis
There were significant differences in explanatory variables between the non-operated (n=47,626) and operated (n=45,549) patients (Table 1). Patients who did not undergo a cholecystectomy were older (64 vs. 54 years, p<0.001), more likely to be male (42.8% vs. 38.0%, p<0.001), and in particular, had more co-morbidities (Charlson Comorbidity Score of 5 or more in 23.5% vs. 8.1%, p<0.001) when compared with patients who had a cholecystectomy. Cholecystectomy was more likely to be performed in high-volume centers compared with medium-or low-volume centers (48.4% vs. 44.3%, p<0.001). Again, age, deprivation, fewer co-morbidities and higher volume centers were independent predictors of cholecystectomy (Table 2). At 1-year following emergency admission for acute cholecystitis (Figure 2), all-cause mortality was significantly higher in the non-operated group when compared with those having a cholecystectomy (12.2% vs. 2.0%, p<0.001). When these were separated into gallbladder related and non-gallbladder related deaths at 1-year, patients who did not have a cholecystectomy had more gallbladder related deaths than those who did have a cholecystectomy (3.3% vs. 0.7%, P o s t -P r i n t p<0.001), however, gallbladder-related deaths were still significantly lower than all other causes of death in the non-operated group (3.3% vs. 8.9%, p<0.001). There were significant differences in explanatory variables between the unmatched emergency and interval cholecystectomy groups (Table 3). Patients undergoing emergency cholecystectomy were more likely to be female (63.1% vs. 61.4%, p<0.001), less deprived (IMD score of 4 or more in 36.2% vs. 34.8%, p=0.002), and had more co-morbidities (Charlson Comorbidity Score of 5 or more in 9.7% vs. 7.2%, p<0.001) when compared with interval cholecystectomy. Emergency cholecystectomy was more likely to be performed in high volume hospitals (45.3% vs. 33.5% vs. 19.4%, p<0.001) when compared with medium-and low-volume hospitals (Figure 3). Similarly, emergency cholecystectomy was more likely to be performed in teaching (38.3% vs. 35.0%, p=0.001) compared with non-teaching hospitals. The total hospital admission time in the year following index emergency admission for cholecystitis was significantly longer for those having emergency cholecystectomy compared with interval cholecystectomy (9 days (IQR 5-18) vs. 8 P o s t -P r i n t days (IQR 4-15), p<0.001). All-cause mortality following index emergency cholecystectomy was higher when compared at 1-year (3.8% vs. 1.0%, p<0.001).

Emergency vs. interval cholecystectomy
A matched one-to-one dataset was created of 15,675 patients undergoing emergency cholecystectomy and 15,675 undergoing interval cholecystectomy. Patient age, gender, ethnicity, IMD scores, and Charlson comorbidity scores were similar between these matched groups (Supplementary Table 1 (Table 4) demonstrated significantly higher risk of all-cause mortality at 1-year following emergency cholecystectomy when compared with interval cholecystectomy (OR 3.3, 95% CI 2.7-3.9, p<0.001).

Discussion
This population-based study demonstrates that 51% of patients in England did not undergo a cholecystectomy within the first year following an index emergency admission with acute cholecystitis. Even when followed up for nearly 5-years, 47% of patients did not undergo a cholecystectomy. Non-operative management of acute cholecystitis is influenced by patient factors (age and co-morbidities) and hospital factors (low and medium volume centers). This nonoperated group had a higher 1-year all-cause and non-gallbladder related mortality consistent with the hypothesis they are a frail patient group. In the 12 months following the index admission, Cholecystectomy remains the treatment of choice for patients with symptomatic gallstones 24,25 despite natural history studies suggesting recurrent pain may only affect half of initially P o s t -P r i n t symptomatic people. 17 A recent systematic review identified only two randomized controlled trials from Norway comparing the clinical and cost effectiveness of cholecystectomy with conservative management in people presenting with uncomplicated symptomatic gallstones or cholecystitis. 26 Fifty-five percent of this heterogeneous cohort randomized to conservative treatment did not undergo a cholecystectomy during the 14-year follow-up period. This trend favoring continued non-operative management is similar to that observed in the current study.
Most of the surgical procedures in this systematic review were performed during the first 5 years and virtually no operations occurred after 5 years. 26 The current data seems to suggest more than 95% of cholecystectomies are performed in the first year in the NHS.
In the current study, patients who did not progress to cholecystectomy were older, with more comorbidities. This may simply reflect that either emergency or interval cholecystectomy was deemed inappropriate due to fitness for surgery. A key limitation of the current study is the lack of quality of life data or patient reported outcomes and it is unclear if these non-operated patients suffered from recurrent symptoms or complications. However, total hospital admission time for biliary causes was actually lower in the non-operated group over the 12 months following the index admission (7 days vs 8 days p<0.001). Natural history studies 17, 27 and more recent observational and population-based studies, 18, 28 suggest nearly half of initially symptomatic patients were relieved of symptoms within the first year of an observation period.
Clearly, there is a group of patients with recurrent symptoms or complications following the index admission with acute cholecystitis. Many clinical trials have concentrated on the timing of cholecystectomy. Index emergency cholecystectomy is reportedly associated with less P o s t -P r i n t gallbladder-specific morbidity, 29-33 a shorter total length of hospital stay, [30][31][32][33][34][35][36][37][38][39][40][41][42][43] lower hospital costs, 30,32,34,40 and fewer lost work days, 30,35,39,44 without increased operative mortality when compared with interval cholecystectomy. 30, 36-39, 41, 45 In contrast, the current population based study, demonstrated significantly lower total hospital admission time in the interval group and mortality was three-fold higher following index emergency cholecystectomy when compared with interval cholecystectomy.
There is an issue of external validity in these non-pragmatic randomized clinical trials. 46,47 With respect to the current analysis, it should be realized both the patient populations and healthcare systems studied in the clinical trials are different. Patients having index emergency cholecystectomy were older with more co-morbidities than those that had interval cholecystectomy. Furthermore, they have waited a median of three days before an operation suggesting they may have failed antibiotics and conservative treatment. There are also likely to be system processes including the availability of resources in specialty units, and a variation in clinical decision making by subspecialty surgeons with expertise in upper gastrointestinal or hepatobiliary surgery. 48 The results presented here support a strategy to improve pathways to enable more efficient interval cholecystectomy. There are valid concerns of waiting for an interval cholecystectomy following an acute admission with cholecystitis. Some of these patients are likely to suffer gallstone-related complications, and recurrent admissions. [3][4][5][6][7] However, the current study suggests nearly 80% of the patients in the interval group do not have another hospital admission prior to cholecystectomy, and currently wait a median of 103 days.

P o s t -P r i n t
Although HES provides robust datasets on medical diagnoses, interventions, outcomes and explanatory variables, careful analysis and interpretation is required due to the retrospective, observational nature of data collection. HES data do not include detailed information on patient physiology, investigative findings, disease severity or patient reported outcomes. Patients admitted as an emergency have a higher risk of poorer outcomes. HES data unfortunately lacks the availability of comprehensive patient, disease or operative data to allow for risk stratification.
Factors such as operative time, and case difficulty, can influence the outcomes studied here.
Emergency cholecystectomies can be complex operative cases, with longer operative times, 35,38,39,49,50 and greater intra-operative blood loss, 35,37 when compared with interval cholecystectomies. 35,37,50 HES does not provide information on the impact on quality of life, patient attitudes and patient satisfaction. In other studies, patient preference was predictably for emergency cholecystectomy, 41 but analgesic use and pain score were not statistically significant when emergency and interval cholecystectomy were compared. 37,38,43,50,51 There is also some evidence that emergency cholecystectomy may cause more pain in the early postoperative period. 35 The current study did not review any detail on temporizing interventions such as percutaneous gallbladder drainage procedures, or the influence of disease severity in the surgical decisionmaking process. Whilst radiological drainage procedures are coded for in the HES database, it is unclear whether these are for the drainage of collections for unrelated reasons, or specifically cholecystostomy drains. These factors together with patient choice can play a large part in influencing case selection for and against index emergency cholecystectomy, and consequently operative outcomes. Future guidelines supporting index emergency cholecystectomy should therefore consider these factors, together with the availability of and access to appropriate radiological, endoscopic and specialist biliary resources.

Conclusion
Strict oversight needs to be considered when financially incentivizing index emergency cholecystectomy. The current study would suggest default policies of emergency cholecystectomy for all patients during an index emergency admission with acute cholecystitis may risk overtreatment, when watchful waiting or judicious interval cholecystectomy may be more appropriate. 26 P o s t -P r i n t   P o s t -P r i n t P o s t -P r i n t P o s t -P r i n t P o s t -P r i n t P o s t -P r i n t