Data were extracted for 219,468 admissions to 159 adult, general critical care units–intensive care units (ICUs) and combined intensive care/high-dependency units–from the Intensive Care National Audit & Research Centre (ICNARC) Case Mix Programme Database (CMPD) for the period of December 1995 to June 2003. Data were collected prospectively and abstracted locally by trained data collectors according to precise rules and definitions. The data underwent extensive validation before incorporation into the CMPD. The process of data validation and cleaning has been described in detail previously 9.

The primary and secondary reasons for admission to the critical care unit are coded in the CMPD by means of a hierarchical coding system, the ICNARC Coding Method 10. Admissions were identified as having SAP if their primary reason for admission was recorded as 'acute pancreatitis,' 'chronic pancreatitis,' or 'infective pancreatitis.' Admissions transferred in from another critical care unit were excluded. APACHE II exclusions (age less than 16 years, length of stay less than 8 hours) were applied when calculating APACHE II scores only 11. Readmissions to the critical care unit within the same hospital stay were excluded from calculation of ultimate hospital mortality and length of stay in hospital.

Age at admission and gender were extracted. Surgical status was defined by the source of admission to the critical care unit and (for admissions from theatre) by the National Confidential Enquiry into Patient Outcome and Death classification of surgery as elective (including scheduled) or emergency (including urgent). Three serious conditions from the past medical history were extracted using data collected as part of the chronic health evaluation section of the APACHE II score 11. These were biopsy-proven cirrhosis documented at any time prior to or at admission to the critical care unit; hepatic encephalopathy (grade 1 or greater) occurring within the six months prior to admission; and steroid treatment, defined as more than or equal to 3 mg/kg prednisolone or equivalent daily for the six months prior to admission.

Mechanical ventilation during the first 24 hours following admission to the unit was identified by the recording of a ventilated respiratory rate. The following physiological parameters, selected a priori, were extracted from the data collected during the first 24 hours in the ICU: lowest arterial oxygen tension (PaO₂), lowest PaO₂/FIO₂(PaO₂/fractional inspired oxygen) gradient, lowest arterial pH, highest serum urea, highest serum creatinine, highest serum glucose, highest total serum bilirubin, lowest total serum calcium, lowest ionised serum calcium, lowest serum albumin, highest white blood cell count, and lowest platelet count. Overall acute severity of illness was summarised with the APACHE II Acute Physiology Score (APS) and the APACHE II score.

Seven of the eight modified Glasgow criteria for assessing severity of pancreatitis 8 were matched based on data available from the first 24 hours following admission to the unit. These were age older than 55 years, white blood cell count greater than 15 × 10⁹/l, serum glucose greater than 10 mmol/l, serum urea greater than 16 mmol/l, PaO₂ less than 8 kPa, serum albumin less than 32 g/l, and total serum calcium less than 2 mmol/l. The remaining criterion that could not be matched was lactate dehydrogenase more than 600 IU/l (lactate dehydrogenase is not recorded in the CMPD). According to the original definitions, these parameters should be assessed during a 48-hour period. However, data were available for only 24 hours.

Mortality data were collected at discharge from the admitting critical care unit (critical care unit mortality) and at ultimate discharge from an acute hospital (ultimate hospital mortality).

Length of stay in the critical care unit was calculated (in fraction of days) from the date/time of admission to and date/time of discharge from or death in the critical care unit. For critical care unit survivors, post-discharge length of stay in hospital was calculated (in days) from the date of discharge from the critical care unit and date of ultimate discharge from an acute hospital or death in hospital. Transfers out to another ICU were identified as critical care unit survivors whose destination following discharge from the unit was an ICU in the same or another hospital. Readmissions to the critical care unit in the same hospital stay were identified from the postcode, date of birth, and gender and were confirmed by the participating units.

The case mix, outcome, and activity of admissions with SAP were described using the variables defined above with numbers and percentages (for categorical variables) or medians and interquartile ranges (for continuous variables).

Ultimate hospital mortality was also reported by the number of modified Glasgow criteria and the APACHE II score. The ability of the modified Glasgow criteria to discriminate survivors from non-survivors was explored with receiver operating characteristic (ROC) curves. The ROC curve for the number of modified Glasgow criteria was compared to those for the APACHE II score and for the 12 individual physiological parameters listed above.

The effect of case mix on length of stay in the critical care unit was explored by comparing median lengths of stay by quartiles of APACHE II score and survival status. All analyses were performed using Stata 8.2 (StataCorp LP, College Station, TX, USA).

In total, 2,677 admissions were identified as SAP, representing 1.2% of all admissions in the CMPD. Of these, 2,376 (88.8%) were coded as acute pancreatitis, 159 (5.9%) as infective pancreatitis, and 142 (5.3%) as chronic pancreatitis. All were considered to represent SAP because their pancreatitis was recorded as the primary reason for admission to the critical care unit. Admissions coded with chronic pancreatitis were assumed to have a severe acute exacerbation of chronic pancreatitis.

The case mix of admissions with SAP is summarised in Table 1. The age and gender distributions were comparable to those of all admissions in the CMPD 9; however, as one would expect, the majority of admissions (84%) were non-surgical. Serious conditions in the past medical history were rare, and only 8, 13, and 27 admissions met the definitions of hepatic encephalopathy, biopsy-proven cirrhosis, and steroid treatment, respectively. The median APACHE II APS (for eligible admissions) was 13, and the median APACHE II score was 17. Between 14% and 72% of admissions met each of the seven modified Glasgow criteria.

Of all admissions with SAP, 31% died in the admitting critical care unit (Table 2). Of all patients with SAP admitted to a critical care unit (excluding readmissions of the same patient within the same hospital stay), 42% died before ultimate discharge from an acute hospital (Table 2). Increasing numbers of modified Glasgow criteria were associated with increasing mortality, rising from 13% for admissions meeting no modified Glasgow criteria to 100% for the five admissions meeting all seven modified Glasgow criteria (Figure 1). Ultimate hospital mortality also increased with increasing APACHE II score, rising from 6% for admissions with an APACHE II score below 10 to 100% for the 13 admissions with an APACHE II score of 40 or higher (Figure 2).

Areas under the ROC curves (aROCs) for the number of modified Glasgow criteria, APACHE II score, and 12 physiological parameters are shown in Table 3. The number of modified Glasgow criteria showed moderate ability to discriminate between hospital survivors and non-survivors (aROC = 0.669), but this was exceeded by the APACHE II score (aROC = 0.804) and four of the individual physiological parameters: highest serum urea, lowest arterial pH, highest serum creatinine, and lowest serum albumin. The ROC curves for the number of modified Glasgow criteria, the APACHE II score, and these four physiological parameters are shown in Figure 3.

Overall median length of stay in the critical care unit was similar for survivors and non-survivors (3.7 versus 3.8 days) (Table 2). However, median length of stay increased with increasing severity of illness for survivors and decreased with increasing severity of illness for non-survivors (Figure 4). Length of stay in hospital after discharge from the critical care unit was also similar for hospital survivors and non-survivors (median 15 versus 14 days) (Table 2), although the lengths of stay of non-survivors were more variable (interquartile range 5 to 38 days compared to 9 to 30 days for survivors). Five percent of all admissions with SAP were transferred out to another ICU, and 5% of admissions represented a readmission of a patient previously admitted during the same hospital stay (Table 2).