Gallbladder perforation (GBP) is an uncommon complication of acute cholecystitis, and previous large series have reported an incidence of approximately 2–11 %. It is associated with high morbidity and mortality rates, and unfortunately, its early detection can be difficult [1–3].

Delay in diagnosis seems to be the major cause of its high morbidity and mortality, and no studies have investigated whether or not risk prediction scores can be helpful in identifying high-risk patients [4].

The aim of this study is to analyse all factors associated with morbidity and mortality, and compare the accuracy of preoperative risk prediction scores.

Methods

From 2002 to 2012, 1,033 patients were submitted to cholecystectomy for acute cholecystitis at our Institution. Only patients with GBP were considered for this study.

GBP was defined as the presence of intraoperative breach of gallbladder wall, postoperatively confirmed by the pathologist.

Patients with previous percutaneous cholecystostomy were not considered for this study; however, our policy is to avoid percutaneous drainage for GBP, as we believe surgical approach is more indicated in these cases.

Preoperative data including demographics, BMI > 35, comorbidities, preoperative sepsis, preoperative albumin, clinical assessment, type and length of surgery, postoperative length of stay, morbidity and mortality were recorded.

Sepsis was divided in four different grades according to the American College of Chest Physicians/Society of Critical Care Medicine (ACCP/SCCM) definition: grade 1: systemic inflammatory response syndrome (SIRS); grade 2: sepsis; grade 3: severe sepsis and grade 4: septic shock [5].

Morbidity was defined as an unexpected event within 30 days after surgical intervention or during the same hospital admission which was harmful for the patient’s health and required a change of therapeutic strategy. Mortality was defined as any death within 30 days after surgical intervention or during the same hospital admission.

American Society of Anaesthesiology (ASA) [6], physiological & operative severity score for the enumeration of mortality and morbidity (POSSUM) [7] and acute physiology and chronic health evaluation II (APACHE II) [8] scores were calculated.

Briefly, ASA score is a six-category physical status classification system based on patient’s overall physical health; POSSUM has two parts, which include assessment of physiological parameters and operative scores; we decide to include physiological parameters only as part of the preoperative evaluation (age; cardiac status; pulse rate; systolic blood pressure; respiratory status; Glasgow Coma Scale score; serum concentrations of urea, potassium and sodium; haemoglobin concentration; white blood cell count and findings on electrocardiography). The physiological parameters are taken at the time of surgery. The APACHE II classification includes 12 physiological measures (temperature, mean arterial pressure, heart rate, respiratory rate, oxygenation, arterial pH, serum sodium, serum potassium, serum creatinine, haematocrit, white blood cell count and Glasgow Coma Scale score), age and severe chronic health problems.

Characteristics of patients and their outcomes were determined using statistical analysis. Discrete, categorical variables were analysed using Chi-squared or Fisher’s exact test when necessary and appropriate. For means in case of continuous numeric data, we used the independent samples t test and the Mann–Whitney U test, respectively, for data normally and nonnormally distributed; the data were previously tested for normality by the Kolmogorov–Smirnov test. All statistical tests were 2-sided. A P value <0.05 was considered statistically significant.

Neimeier’s classification was also used (Type I: acute free perforation into the peritoneal cavity; Type II: subacute perforation with pericholecystic abscess and Type III: chronic perforation with cholecystoenteric fistula) [9].

The threshold of statistical significance following univariate analysis was set at P < 0.1. Multivariate analysis of statistically significant variables was performed using binary logistic regression.

Receiver operator characteristic (ROC) curves were analysed to examine sensitivity and specificity of each ASA, POSSUM and APACHE II scores. The area under the curve (AUC) was compared to measure the ability of the scores.

All statistical analyses were done with the statistical software package SPSS (version 19; SPSS Inc., Chicago, IL, USA).

Results

From 2002 to 2012, 137 patients underwent emergency cholecystectomy for acute cholecystitis with GBP. Median age was 76 (36–99) years, median number of days from admission to surgery was 1.4 (0–22) days, 24.1 % of patients received laparoscopic surgery, and 65 (47 %) patients had grade 1–4 sepses at the time of the operation. Demographic and clinical factors are reported in Table 1. Seventy-nine (57.7 %) patients had at least one complication. Mortality rate was 9.5 %.

Table 1 Demographics and clinical characteristics
Full size table

Morbidity

At univariate analysis, age, cardiac history, oral anticoagulation, preoperative sepsis and number of days between admission and surgery were associated with postoperative complications. Laparoscopic operation and preoperative albumin were also inversely associated with complications (Table 2). At multivariate analysis, low preoperative albumin, open surgery and preoperative sepsis were independently associated with complications (Table 3).

Table 2 Univariate analysis of factors associated with postoperative complications
Full size table
Table 3 Multivariate analysis of factors independently associated with complications
Full size table

Mortality

At univariate analysis, age, interval to surgery and preoperative sepsis were significantly associated with mortality (Table 4). At multivariate analysis, preoperative sepsis was the only factor independently associated with hospital mortality (Table 5).

Table 4 Univariate analysis of factors associated with mortality
Full size table
Table 5 Multivariate analysis of factors independently associated with mortality
Full size table

Scores

ASA, POSSUM and APACHE II scores were significantly associated with postoperative complications and mortality (Tables 3, 5).

APACHE II score showed the best accuracy in both predicting morbidity (Fig. 1, AUC: APACHE II = 0.729 vs. ASA = 0.635 and POSSUM = 0.721) and mortality (Fig. 2, AUC: APACHE II 0.799 vs. ASA 0.695 and POSSUM 0.782).

Fig. 1
figure 1

ROC curve for postoperative complications

Full size image
Fig. 2
figure 2

ROC curve for postoperative mortality

Full size image

Discussion

GBP is a life threatening complication of acute cholecystitis. Despite advances in early detection, the morbidity and mortality rates are still very high and have not dramatically changed over the years. Since the difficulties in diagnosis cause delay in treatment, higher morbidity and mortality rates are often encountered. Glenn and Moore originally reported a 42 % mortality rate following GBP [10], although more recent studies reported a 12–16 % rate [11].

In the present study, the morbidity and mortality rates were 57.7 and 9.5 %, respectively. Very little has been previously published about the outcome of these groups of patients [12]: we present the largest series ever published trying to detect those patients at risk of developing complications and dying during the hospital admission.

Three factors showed significant association with morbidity: preoperative albumin, laparoscopic surgery and preoperative sepsis/SIRS. Preoperative albumin is a very well-known factor associated with morbidity in most of abdominal operations, including elective surgery [13]. Low serum albumin has been historically linked to nutritional status; however, the relationship of albumin to inflammation was not initially recognized, and it is now known that inflammation is the most important variable that affects hepatic protein metabolism, and that inflammatory activity diminishes the ability of the patient to respond well to a second hit [14, 15]; however, correction of hypoalbuminemia to prevent postoperative complications is not recommended, making this finding not very useful [16].

The association between laparoscopic approach and lower morbidity could have been biased by several factors: firstly, laparoscopic cholecystectomy (LC) is associated with fewer complications when compared to open surgery [17], but there are no studies validating these results in GBP patients; secondly, laparoscopic approach was probably performed in easier cases and finally, some patients who would intuitively benefit the most from a minimally invasive approach did not undergo LC. The reasons might be that some of our general surgeons were not skilful laparoscopic surgeons. Many patients were operated as emergency open cholecystectomy during on-call hours by general surgeons, which explains our relatively low rate of laparoscopic cholecystectomies. It is obvious that hepatobiliary surgeons with good experience of laparoscopic technique are most skilled to remove perforated gallbladder using laparoscope, but they are not always available during on-call hours.

Unfortunately, these are limitations of our retrospective study, but this is what probably happens in most hospitals of the world.

Preoperative sepsis/SIRS was certainly the determinant factor associated with mortality and morbidity. Its role as a risk factor for postoperative complications and death has been already described in abdominal surgery, including hepatobiliary [18–21].

Differently than hypoalbuminemia and type of surgery, preoperative sepsis is the only factor that can change our therapeutic strategy, for instance administering broad spectrum antibiotics from the start and admitting these patients in a high-dependency unit. Most of the recent literature has been focusing on the importance of early surgery in acute cholecystitis patients [22], which is something we were expecting to be associated with good outcome in our series. However, interval to surgery lost its significancy into the multivariate analysis, probably due to the different population we evaluated: in case of GBP, the clinical presentation (sepsis) is more important than timing.

With regard to risk prediction of morbidity and mortality, APACHE II score was superior to ASA and POSSUM. However, the final accuracy of these tests was fair (AUC < 0.8) and we would not recommend using these tests to stratify patients’ GBP at risk of postoperative complications; in our series more than 50 % of the patients had at least one complication, and using these scores is unlikely to change their management.

It could be argued that early identification of patients with GBP might improve the final outcome; unfortunately, preoperative diagnosis remains the most challenging problem [23]; abdominal ultrasound has fair sensitivity [24], and no clinical criteria can be used to differentiate GBP from acute cholecystitis. Some authors have demonstrated that CT scan can predict GBP in up to 81 % of patients thought the visualization of GB wall defects, gas and ascites, but these are retrospective findings and only when a CT was performed to confirm US findings [25, 26]. In our series, most of the patients were not submitted to abdominal CT scan and it would be very difficult to justify its routine use in patients with acute cholecystitis within a busy emergency department.

Previous studies have shown that age >60, male gender and systemic disorders are associated with GBP [27, 28], but they really are generic factors.

The results of our study should not be misinterpreted concluding that sicker patients have worse outcome. In fact, comorbidities lost their significancy in multivariate analysis, meaning that sepsis is the main determinant regardless of patient’s status prior to surgery and that is confirmed by the fair accuracy of risk scores that were tested.

Until preoperative detection of GBP will not be improved, whether or not percutaneous drainage should be preferred in patients with severe sepsis will remain matter of debate.

We can only recommend that those patients with intraoperative finding of GBP should be considered at very high risk for complications and mortality, and therefore broad spectrum antibiotics and monitoring in a high-dependency unit should be immediately started.