Relationship between intra‐abdominal hypertension, outcome and the revised Atlanta and determinant‐based classifications in acute pancreatitis

Background The aim of this study was to analyse the relationship between intra‐abdominal hypertension (IAH) and severity of acute pancreatitis (AP) measured by the revised Atlanta classification (RAC) and determinant‐based classification (DBC). Secondary objectives were to assess IAH as a predictor of morbidity and mortality in the ICU. Methods This prospective international observational study included patients admitted to the ICU with AP and at least one organ failure. Information was collected on demographics, severity scores at admission using RAC and DBC, organ failure, mechanical ventilation, continuous renal replacement therapy (CRRT), surgery and mortality. Maximum intra‐abdominal pressure (IAP) during ICU stay was used for analysis. Results Some 374 patients were included. The hospital mortality rate was 28·9 per cent. IAP was measured in 301 patients (80·5 per cent), of whom 274 (91·0 per cent) had IAH and 103 (34·2 per cent) acute compartment syndrome. A higher IAH grade was more likely in patients with severe AP (42 per cent for grade I versus 84 per cent for grade IV) and acute critical pancreatitis (9 versus 25 per cent; P = 0·001). Compared with grade I IAH, patients with grade IV had more infected necrosis (16 versus 28 per cent; P = 0·005), need for surgery (27 versus 50 per cent; P = 0·006), mechanical ventilation (53 versus 84 per cent; P = 0·007) and requirement for CRRT (22 versus 66 per cent; P < 0·001). IAH predicted shock (area under receiver operating characteristic (ROC) curve (AUC) 0·79, 95 per cent c.i. 0·73 to 0·84), respiratory failure (AUC 0·82, 0·77 to 0·87), renal failure (AUC 0·93, 0·89 to 0·96) and mortality (AUC 0·89, 0·86 to 0·93). Conclusion IAH was associated with severity of AP classified according to both RAC and DBC systems. IAP grade can predict outcome of AP during ICU stay.


Introduction
Intra-abdominal hypertension (IAH) is defined as a repeated pathological intra-abdominal pressure (IAP) increase of 12 mmHg or more. When IAP is above 20 mmHg (with or without an abdominal perfusion pressure below 60 mmHg) and associated with organ failure (mainly cardiovascular, respiratory or renal dysfunction), it is called abdominal compartment syndrome (ACS). According to its severity, IAH is classified in four categories 1 -3 ( Fig. 1).
IAH and ACS are common events in acute pancreatitis (AP), affecting up to 60-80 per cent of patients in some studies 4,5 . The proportion of patients thought to have IAH depends on where the threshold is set. With an IAP greater than 15 mmHg, over 75 per cent of patients might be classified as having IAH 4,5 , often within the first 12 h of their apparent illness 6,7 . Some patients with AP may develop ACS (25-56 per cent) 4 -8 , associated with a high mortality rate (50-75 per cent) 4 -8 . IAH in AP with organ failure is generally thought to reflect visceral oedema due to the severity of the inflammatory process, potentially compounded by aggressive fluid resuscitation (volume overload).
Recently, two classification systems have been developed for patients with AP to predict mortality risk: the revised Atlanta classification (RAC) 9 and the determinant-based classification (DBC) 10 . These two systems are useful to characterize case mix and to compare the effectiveness of treatments between centres. Initial validation studies 11,12 showed that the two classifications were equally accurate in predicting mortality, whereas other systems used to assist clinicians in the evolution of AP during an ICU stay have low accuracy 13,14 . IAH may be useful to predict local pancreatic and systemic complications 8,15 -18 . This study sought to examine the relationship between IAP and the RAC and DBC systems. Secondary objectives were to assess morbidity and mortality associated with IAH in patients with AP.

Methods
This was a prospective observational multicentre international cohort study conducted over a 1-year period. Following the consensus conference on the management of AP under the auspices of the Spanish Intensive Care Society (SEMICYUC) 3 , invitations were sent to all international ICUs that contributed to the Epidemiology of Acute Pancreatitis in Intensive Care Medicine (EPAMI) study. In addition, several ICUs in South America were invited to participate. Participating centres are shown in Table S1 (supporting information).
Study inclusion criteria were: patient aged over 18 years, admitted to an ICU with a diagnosis of AP and at least a single organ failure. The following definitions were used: AP diagnosed by at least two of the following three criteria -upper abdominal pain, amylase or lipase levels raised at least threefold with respect to laboratory upper limits of normal, appropriate imaging findings; organ failure -RAC and DBC definitions, adapted to intensive care, using SEMICYUC consensus conference criteria 3 ; shock -systolic arterial pressure below 90 mmHg or a reduction of 40 mmHg in basal systolic arterial pressure, with tissue hypoperfusion signs where lactate concentration was greater than 3 mmol/l and central venous oxygen saturation was less than 70 per cent 19 ; respiratory failure -either basal arterial partial pressure of oxygen (PaO 2 ) below 60 mmHg, or PaO 2 or fraction of inspired oxygen (FiO 2 ) of less than 300 mmHg with supplementary oxygen; acute renal failure -increase in serum creatinine to more than twice the upper limit of normal and/or less than 0⋅5 ml per kg per h for more than 12 h according to urinary output criteria for the Acute Kidney Injury Network (AKIN) classification; IAH -a repeated pathological IAP increase of 12 mmHg or above; and ACS -when IAP was above 20 mmHg with or without an abdominal perfusion pressure below 60 mmHg and associated with an organ failure that was cardiovascular, respiratory or renal 1,20 . There were no exclusion criteria. Study recruitment was from 1 January to 31 December 2013. The study was approved by Donostia -San Sebastián University Hospital ethics committees and endorsed at other centres. Each patient provided signed informed consent before entering the study. All data were anonymized.
Data were collected for all patients relating to: age, sex, aetiology, shock, respiratory failure, acute renal failure, local complications, two widely used ICU scores to predict mortality (Acute Physiology And Chronic Health Evaluation (APACHE) II and Sequential Organ Failure Assessment (SOFA) in first 24 h), use of mechanical ventilation, continuous renal replacement therapy (CRRT), surgery, ICU length of stay (LOS), overall hospital LOS and hospital mortality.
IAP was measured via the urinary bladder at end expiration in the completely supine position after ensuring that abdominal muscle contractions were absent and with the transducer zeroed at the level of the mid-axillary line, according to the updated consensus definition from the World Society of the Abdominal Compartment Syndrome 1,20 . IAP was measured every 6 h daily, and maximum IAP value obtained during the ICU stay was the one analysed. Owing to design issues, the day on which value occurred was not collected. Maximum IAP was graded according to the same consensus (Fig. 1).

Statistical analysis
Quantitative variables with a normal distribution were expressed as mean(s.d.) and those with a non-parametric distribution as median (i.q.r.) values. Categorical data were expressed as frequencies and percentages.   Table 1. IAP was measured in 301 patients (80⋅5 per cent). Mean maximum IAP was 19⋅2(5⋅8) mmHg. Of these patients, 274 (91⋅0 per cent) developed IAH, among whom 110 (36⋅5 per cent) were considered at risk of developing ACS because they had an IAP above 20 mmHg. ACS developed in 103 patients (34⋅2 per cent), of whom nine (8⋅7 per cent) underwent decompressive laparotomy; three had an IAP of 21-25 mmHg and six an IAP above 25 mmHg. Seven of these nine patients died.
There was a direct significant relationship between the development of infected pancreatic necrosis, need for surgery and graded IAH (P = 0⋅006). A direct relationship existed between ICU and hospital LOS and graded IAH (P < 0⋅001). Enteral nutrition was used most frequently in patients with grade I (40 per cent) and II (23⋅9 per cent) IAP than in those in grade III or IV (both 19 per cent) (P = 0⋅009) ( Table 2).

Relationship between graded intra-abdominal hypertension, revised Atlanta and determinant-based classifications
When RAC was applied, there were 163 patients (43⋅6 per cent) with moderate and 211 (56⋅4 per cent) with severe AP. With the DBC, there were 135 patients (36⋅1 per cent) with moderate forms of AP, 175 (46⋅8 per cent) with severe AP and 64 (17⋅1 per cent) with critical AP. Clinicians measured IAP more frequently in severe and critical forms of AP than in moderate ones (P < 0⋅001). There was a direct relationship between graded IAH and the severity of AP according to the RAC and DBC systems (P = 0⋅001). For example, there were no critical forms of AP according to the DBC in patients with a normal IAP, and 74 per cent of patients with grade III IAH and 84 per cent of patients with grade IV IAH developed a severe form of AP according to the RAC ( Table 3).

Discussion
This study has shown a significant relationship between graded IAH and the severity of AP stratified by the RAC and DBC systems. IAP should be measured in patients with AP as it can predict severity and potentially influence management. The study confirmed the relationships between IAP, organ failure and mortality in the ICU in line with other observations 6,16 -18 . Although one study 17 found that an IAP of 9 mmHg provided the best predictive value of 30-day mortality (sensitivity 86 per cent, specificity 87 per cent; AUC ROC 0⋅91), this very low cut-off point probably reflects the fact that patients in the present study all had AP and at least one organ failure.
Just over one-third of patients in the present study developed ACS, similar to other results ranging from 27 to 45 per cent 15,18,21 . Unlike earlier studies where rates of decompressive laparotomy were high (74-77 per cent) 18,21 , only 8⋅7 per cent of the present cohort diagnosed with ACS were treated in this way. Some patients in the present series were treated with effective medical decompressive methods (nasogastric and rectal tubes, percutaneous drainage of abdominal collections, negative fluid balance with diuretics and renal support), as recommended elsewhere 1 -3 . As the treatment of ACS was not specified by protocol in the study design, participating centres may not have considered that existing data from decompressive laparotomy justified this procedure. In both of the above studies 18,21 , the mortality rate of patients with ACS was high (49 and 54⋅5 per cent respectively). The mortality rate was also high in the present study: 40 per cent for patients with grade III IAH, 66 per cent for patients with grade IV IAH, and seven of nine patients (78 per cent) who had decompressive laparotomy.
This study had limitations. EPAMI was an observational study and its purpose was focused on epidemiology and AP classification. IAP and IAH were analysed as secondary variables. IAP was not measured in all patients, and the timing of the highest IAP may not have been recorded. Despite these shortcomings, IAP was significantly associated with severity of AP according to both the RAC and DBC.
Graded IAP was predictive of outcome in patients with AP during the ICU stay in the present study. Its true value in predicting the development of organ failure and mortality should be investigated in a study designed specifically for this purpose.

Graphical Abstract
The contents of this page will be used as part of the graphical abstract of HTML only. It will not be published as part of main article. Graded intra-abdominal hypertension (IAH) accurately predicts shock, respiratory failure, renal failure and mortality in acute pancreatitis (AP). Graded IAH could be validated as another classification for the severity of AP because of its excellent relationship with the already validated revised Atlanta and determinant-based classifications.