First attack of acute pancreatitis in Sweden 1988 – 2003: incidence, aetiological classification, procedures and mortality – a register study

Background Population-based studies suggest that the incidence of first attack of acute pancreatitis (FAAP) is increasing and that old age is associated with increased mortality. Beacuse nationwide data are limited and information on standardized mortality ratio (SMR) versus age is lacking, we wanted to describe incidence and mortality of first attack acute pancreatitis (FAAP) in Sweden. Methods Hospital discharge data concerning diagnoses and surgical procedures and death certificate data were linked for patients with FAAP in Sweden. Mortality was calculated as case fatality rate (CFR), i.e. deaths per 1000 patients and SMR using age-, gender- and calendar year-specific expected survival estimates, and is given as mean with 95% confidence intervals. Data are presented as median values with 25% and 75% percentiles, means and standard deviations, or proportions. Proportions have been compared using the chi square test, Poisson-regression test or Fisher exact test. Location of two groups of ratio scale variables were compared using independent samples t-test or Mann-Whitney U-test. Results From 1988 through 2003, 43415 patients (23801 men and 19614 women) were admitted for FAAP. Age adjusted incidence rose from 27.0 to 32.0 per 100000 individuals and year. Incidence increased with age for both men and women. At index stay 19.7% of men and 35.4% of women had biliary diagnoses, and 7.1% of men and 2.1% of women alcohol-related diagnoses. Of 10072 patients who underwent cholecystectomy, 7521 (74.7%) did so after index stay within the audit period. With increasing age CFR increased and SMR decreased. For the whole period studied SMR was 11.75 (11.34–12.17) within 90 days of index admission and 2.03 (1.93–2.13) from 91 to 365 days. Alcohol-related diagnoses and young age was associated with increased SMR. Length of stay and SMR decreased significantly during the audit period. Conclusion Incidence of FAAP increased slightly from 1988 to 2003. Incidence increased and SMR declined with increasing patient age. Although the prognosis for patients with FAAP has improved it remains an important health problem. Aetiological classification at index stay and timing of cholecystectomy should be improved.


Background
The diagnosis of acute pancreatitis is based on clinical features (abdominal pain) and elevation of plasma concentrations of pancreatic enzymes, if indicated supplemented by imaging techniques [1][2][3][4]. Gallstone disease is reported to account for approximately half of all cases with acute pancreatitis and alcohol abuse for 20-25% [5] with great variations between regions and over time.
Population-based studies suggest an increasing trend for incidence of first attack acute pancreatitis (FAAP) [5,6] although it is unclear to what extent this might reflect changes in population age. The presumption that old age is a risk factor for mortality [5,7] is based upon case fatality rate (CFR), which does not take into account mortality in the background population.
We have studied incidence, aetiological classification, procedures, and mortality for patients with FAAP in Sweden 1988-2003, using nationwide hospital discharge and death certificate data. Influence of age upon incidence and mortality was considered of particular importance.

Methods
In Sweden patients are identified through a national registration number unique for each resident in Sweden. The Swedish National Board of Health and Welfare's Epidemiology Centre compiles data on individual hospital discharges in the Hospital Discharge Register (Patient administrativa registret, PAR) [8]. Since 1987, the register has included all Swedish hospitals. The record of each hospital stay contains diagnoses at discharge coded according to the Swedish version of the International Classification of Diseases (ICD), from 1987 through 1996 the 9th revision and from 1997 the 10th revision. Hospital stays are classified as emergent or elective. Surgical procedures are classified according to the Swedish version of Classification of Operations 1985 (revised 1988) and Classification of Surgical Procedures 1997. Underlying causes of death were obtained from Death Certificates, which have to be written for each patient after death in Sweden, are coded by the ICD-9 and ICD-10 classifications.
For all records reported to PAR a data control is run [8]. A quality control is made, that codes for different variables and dates have valid values and that compulsory variables (personal identification number, hospital and main diagnosis) are reported. Obviously incorrect data are corrected. In 2003, 0.9% of all diagnoses and 0.5% of acute somatic diagnoses was missing in hospital stays reported.

Data acquisition
For the present study, data from PAR and from death certificate data were retrieved for all patients with a hospital admission from 1988 through 2003 for acute pancreatitis (577A in ICD-9 or K85 in ICD-10). Information on hospital admissions and procedures one year prior to and one year after index admission were collected for all patients.

Definitions
First attack acute pancreatitis (FAAP) is defined as acute pancreatitis, without hospital visit with this diagnosis, during a minimum of one year preceding index admission. The aetiological diagnosis of FAAP refers to primary or secondary diagnoses during index admission; classified as biliary disease, alcohol-related disease or other (none of these) diagnoses. ICD codes in addition to K85.0 or 577A used for aetiological classification and for causes of death are given below: Procedures were classified as biliary procedures or operations on pancreas (codes introduced in 1997 with capital letters). , open or laparoscopic JLW,  5520, -21, 5590, -91, -92, -93, 5599 Statistical analysis Data are presented as median values with 25% and 75% percentiles, means and standard deviations, or proportions. Proportions have been compared using the chi square test, Poisson-regression test or Fisher exact test when appropriate. Location of two groups of ratio scale variables were compared using independent samples t-test or Mann-Whitney U-test. Mortality was calculated as case fatality rate (CFR), i.e. deaths per 1000 patients and standardized mortality ratio (SMR) using age-, gender-and calendar year-specific expected survival estimates from Statistics Sweden [9]. SMR is given as mean with 95% confidence intervals. Calculations were performed with SPSS 15.0 (SPSS Inc. Chicago, IL, USA). A p-value less than 0.05 was considered significant.

Ethics
Ethical approval for the study was obtained from the Regional Research Ethics Committee of the University of Umeå, Sweden (Dnr 06-027M) before commencement of the project.  Table 1 shows number of patients per age group. There is a highly significant (p < 0.001) time-dependent difference in age distribution. This is more evident in Table 2 illustrating age specific incidences, in which the crude overall incidence and incidence adjusted to the Swedish population of 1988-1992 are given. Within each time period, the incidence of first attack acute pancreatitis increased with patients' age. The age adjusted incidence increased significantly for women (p < 0.001) and for men and women together (p < 0.001) but not for men (p > 0.05). For both men and women above the age of 70 years, the adjusted age specific incidence increased significantly (p < 0.001) with time. Considering all patients, the adjusted incidence rose modestly, from 27.0 per 100000 individuals per year 1988-1992 to 32.0 1998-2003. This increase was entirely due to a significant (p < 0.001,) rise in incidence for women, from 21.1 to 31.0 per 100000 per year.
Mortality and cause of death The CFR and SMR for patients in the three aetiological groups declined significantly for each period during the audit, see Table 9. As shown in      (Tables 11, 12 and 13). For all patients together and for the three groups separately, SMR declined with patients' age. This contrasts to case fatality rates (CFR), which increased with patients' age, outlined in Figure 1.
Cause of death for all patients who died within one year of index admission is given in Table 14. According to death certificates, biliary diagnoses were the underlying causes of death for 10.5% of all patients who died within 90 days of index admission, whereas alcohol-related diagnoses accounted for 1.1%, and all other diagnoses for 88.4%. From three to twelve months after index admission 3.7% of all patients who died had biliary or alcohol-related diagnoses as underlying causes of death.

Discussion
In this register study we found a crude incidence of first attack acute pancreatitis (FAAP) of 27-33 per 100000 individuals per year. Biliary aetiology was more common among women than men, whereas the reverse was true for alcohol-related diagnoses. Two thirds of patients with FAAP lacked an aetiological diagnosis at index stay. SMR was raised twelve fold within 90 days of index admission and decreased with patients' age in contrast to CFR, which increased with increasing age.
We have utilized nation-wide data from all patients with FAAP in Sweden 1988-2003. Personal identification numbers make it possible to trace patients over time and to link hospital discharge data and death certificate data.
The clinical features of acute pancreatitis (abdominal SMR (standardized mortality ratio) given as median (confidence intervals 95%), CFR (case fatality rate) is given as deaths per 1000 patients. SMR (standardized mortality ratio) given as median (confidence intervals 95%), CFR (case fatality rate) is given as deaths per 1000 patients. pain) together with elevation of plasma concentrations of pancreatic enzymes are well known to surgeons and physicians taking care of emergency patients, and the diagnostic validity in our audit can be considered high.
As main measure of mortality we used SMR, thereby adjusting for the mortality of the background population. Like most register studies, we have limited clinical information of patients' clinical conditions. Aetiological diagnoses in our audit refer to classification at index discharge, as treatment of acute pancreatitis should be based upon knowledge of aetiology (and disease severity) at index stay.
The Swedish hospital discharge register do not take into account outpatient procedures. This is unlikely to introduce any substantial error in cholecystectomy rate as day-case cholecystectomy was of limited use during the period studied. Of all cholecystectomies done in 2005 12% were done as day-cases. This may be deduced by comparing in-hospital cholecystectomy and over-all cholecystectomy figures for Sweden [8,10]. Some sphincterotomies may also have been performed as ambulatory procedures, and therefore not registered before or after index stay, most likely in the latter part of the audit period.
In analyses of acute pancreatitis it is of importance to differentiate between first and recurrent attack [4,11,12]. We found that 16.9% of all patients with FAAP had a recurrent attack of acute pancreatitis within one year of FAAP. Recurrent pancreatitis was less common (9.4%) for patients with biliary pancreatitis, most likely attributable to the fact that 55.7% of these patients had a definitive treatment (cholecystectomy) within one year of index admission for FAAP. SMR (standardized mortality ratio) given as median (confidence intervals 95%), CFR (case fatality rate) is given as deaths per 1000 patients. SMR (standardized mortality ratio) given as median (confidence intervals 95%), CFR (case fatality rate) is given as deaths per 1000 patients. Incidence of FAAP comparable to the one found in this audit has been described for other non-UK studies [5,13,14]including Sweden [11,12], whereas a lower incidence is reported from UK [5,15] and the Netherlands [16].
Population-based studies [5,6,14] indicate an increasing incidence with time, particularly for women. In the present 16-year audit, incidence increased slightly but significantly for women but not for men. During the last six years of our audit, crude incidence for men and women differed by less than 10%.
Studies reporting age adjusted incidence rates have found that incidence increases with age [13,14,17,18]. We observed a six-fold increase in age standardized incidence from patients 0-49 years of age to patients 80 years and above. For patients above the age of 70 years, the incidence increased significantly during the audit period. One might speculate that this could be due to a true increase of biliary disease in elderly or to a reduction in cholecystectomy rate. We have no data concerning change in gallstone prevalence during the period studied and no specific data for different age groups. There is a falling cholecystectomy rate during the 80s, followed by a noticeable increase after introduction of laparoscopic cholecystectomy in the early 90s [19,20]. However, according to Danish data this increase does not hold for older ages, where the cholecystectomy rate is decreasing [19].
As in previous studies, biliary aetiology was more common in women than in men, whereas the reverse was true for alcohol-related aetiology. The most striking finding in our audit was, however, that 68.4% of all patients with first attack acute pancreatitis lacked an aetiological diagnosis at index stay. This is at variance with the UK guidelines according to which the aetiology of acute pancreatitis should be determined in at least 80% of cases and no more than 20% should be classified as idiopathic [2]. As treatment should be governed by aetiology and severity of disease at index stay, substandard aetiological classification may lead to suboptimal treatment.
Previous studies [3,21] have demonstrated that some 85% of all patients with FAAP might be expected to have moderate or mild pancreatitis. According to UK guidelines [2], definitive treatment at index stay is motivated in mild gallstone pancreatitis, and unless the patient is unfit for surgery this will be by cholecystectomy. However, expectant management with interval cholecystectomy has been considered appropriate for most patients with mild to moderate pancreatitis and an improving clinical course [3]. Patients with mild acute biliary pancreatitis who are poor candidates for surgery and patients with severe gallstone FAAP in combination with cholangitis, jaundice, or a dilated common bile may benefit from (early) ERCP at index stay [2,3]. In the present audit, approximately five per cent of all patients underwent cholecystectomy and an equal percentage ERCP at index stay, whereas the majority of all cholecystectomies were done after index stay. Obviously, a scrutiny of treatment for mild gallstone acute pancreatitis in Sweden is motivated. It has recently been shown that feed-back to surgeons may increase adherence to guidelines in this field [22].
Previous reports [5] describe that the mortality of patients with FAAP increases with age. Also in the present study, CFR increased with patients' age. However, SMR is a better measure of the excess mortality risk caused by the disease studied, as it adjusts for the mortality of background population. We found that SMR decreased with age irrespectively of aetiological diagnosis of FAAP. This seems to indicate that the vulnerability of the pancreas declines with age. In this connection, the finding of Sotoudehmanesh et al. [23] that the risk for post-ERCP pancreatitis is reduced in patients above the age of 60 years is of interest.
In the present study SMR was elevated twelve-fold 0-90 days after admission for first attack acute pancreatitis and doubled 91-365 days after admission, considering all patients together. For patients with biliary or alcoholrelated diagnoses the SMR within 90 days was 5.84 and 58.70, respectively. It has been noted previously that patients with acute pancreatitis of alcohol-related aetiology are more likely than patients with gallstone pancreatitis to develop severe disease [24]. However, similar mortality rates have been reported for acute pancreatitis of different origin [5,25]. These conflicting results are best explained by the fact that the same CFR value is associated with a higher SMR value in younger patients (with alcohol-related acute pancreatitis) compared to older patients (with gallstone associated pancreatitis).
The over-all importance of aetiological factors behind severe FAAP in Sweden is best described by our finding that biliary diagnoses were 10 times more frequent than alcohol-related diagnoses as the underlying cause of death within 90 days.
The need to identify cost-effective strategies to prevent and treat acute pancreatitis has been emphasized [26][27][28] and our study underlines this. In agreement with previous studies [5] we found that CFR (and SMR) decreased in recent years. This observation and the shortening of the length of stay during the audit period seem to indicate that treatment of FAAP in Sweden has improved.

Conclusion
An overall incidence of 33 per 100000 inhabitants per year and a mortality rate twelve times higher than for the background population during the first 90 days after hospital admission make FAAP a health care problem of great importance in Sweden. There is room for improvement of aetiological diagnosis at index stay and for management of acute pancreatitis of biliary origin.