Population-based studies on the epidemiology of primary sclerosing cholangitis (PSC) are sparse. The aim of the present study was to investigate prevalence and temporal trends in the incidence of PSC 1992-2005 in an adult population in Västra Götaland, a region in southern Sweden with a defined population of about 1.5 million. Patients with PSC aged 18 years or above were identified through a computerized search for diagnostic codes. A total number of 199 incident cases fulfilling Mayo criteria for PSC were identified through retrospective validation of clinical records. Temporal trends in the incidence of PSC were investigated by Poisson regression and expressed as average annual percent change (AAPC) with a 95% confidence interval (CI). The point prevalence of PSC on December 31, 2005, was 16.2/100,000 in the total adult population (men, 23.7/100,000; women, 8.9/100,000). The annual incidence was 1.22/100,000 in the total adult population (men, 1.78/100,000; women, 0.69/100,000). The overall incidence rate of PSC increased significantly over the investigation period (AAPC 3.06, 95% CI 0.01-6.20). Stratified analysis revealed significantly increasing trends for inflammatory bowel disease (IBD)–associated PSC (AAPC 7.01, 95% CI 0.24-14.24) and large duct PSC (AAPC 6.32, 95% CI 0.03-13.02) in women and for PSC without IBD (AAPC 9.69, 95% CI 0.82-19.33) and small duct PSC (AAPC 17.88, 95% CI 0.95-40.25) in men. Conclusion: This is the first study to report a significantly increasing trend in the incidence of PSC. The prevalence of PSC at the end of the study period is the highest reported to date. This implies that the medical burden of PSC may be higher than estimated previously. (HEPATOLOGY 2010;)
Primary sclerosing cholangitis (PSC) is a chronic cholestatic liver disease characterized by inflammation and fibrosis of the biliary tree. Biliary cirrhosis and liver failure are common complications during long-term follow-up. 1-3 Studies on the natural history of PSC have reported a median liver transplant–free survival between 12 and 18 years. 4-7 PSC usually presents in early adulthood and is more common in men than in women. 4, 8 Inflammatory bowel disease (IBD) is present in approximately 70% of cases in most studies. 4, 6, 8, 9
The epidemiology of PSC has mainly been investigated in case series from referral centers. This introduces a risk for selection bias and does not allow estimates of prevalence and incidence rates. Current knowledge on incidence and prevalence of PSC is derived from small population-based studies 10-13 reporting incidence rates of 0.9-1.3 and prevalence rates of 8.5-13.6 per 100,000 person-years. One recent study from the United Kingdom investigated the incidence and prevalence of PSC in the General Practice Research Database, reporting a considerably lower incidence and prevalence compared with previous studies. 14 The need for larger population-based studies has been emphasized by several investigators. 11, 12
Västra Götaland is a regional health care unit in Sweden wherein one university hospital and seven regional hospitals together serve a defined population of about 1.5 million. Exact age- and sex-stratified population statistics are available. All liver transplantations in the region are performed at the university hospital, and no patients with PSC are referred to hospitals outside the region. This allows complete retrieval of cases with PSC and calculation of age- and sex-standardized incidence rates. The aim of the present study was to perform a population-based study on the incidence and prevalence of PSC in a sufficiently large population over a period long enough to allow analysis of temporal trends in the incidence of PSC. The adult population of Västra Götaland from 1992 to 2005 was therefore considered an appropriate study population.
Patients and Methods
Västra Götaland is a regional health care unit in southwestern Sweden for which detailed annual age- and sex-stratified population statistics are available. The mean number of inhabitants in the region was 1,492,000 during the period 1992-2005. Gothenburg is the largest city in the region and had a mean population of 788,000 during the study period. There is one university hospital situated in Gothenburg (Sahlgrenska University Hospital), and seven regional hospitals are spread throughout the region. Referral of patients from regional hospitals or from hospitals outside the region to the university hospital occurs in selected cases for clinical second opinion, advanced diagnostic or therapeutic procedures, and/or liver transplantation. No patients are referred out of the region.
Patients with PSC during the study period 1992-2005 were identified from inpatient and outpatient registers at all departments of internal medicine and surgery at all hospitals in the region using a computerized search for relevant codes according to the International Classification of Diseases, ninth and tenth revision (codes 576 and K830, respectively). The computerized search was extended to the end of 2006 (1 year beyond the end of the study period) so as to include patients with an uncertain diagnosis at the end of the study period. Only subjects aged ≥18 years were considered; no diagnostic searches were performed in pediatric clinics. Clinical, pathological, and radiological records were reviewed in all cases and were validated according to a standard protocol. Information collected in the protocol included sex, date of birth, date of diagnosis, date of death, date of loss of follow-up, and diagnostic workup, including all cholangiographic examinations, liver biopsies, and results of relevant blood tests. The diagnosis of large duct PSC was defined using Mayo criteria: (1) laboratory findings consistent with cholestasis including elevated serum alkaline phosphatase; (2) characteristic radiological findings proven by endoscopic, percutaneous, or perioperative cholangiography or magnetic resonance cholangiopancreaticography (MRCP); and (3) no evidence for secondary sclerosing cholangitis. 11 Cases were defined as small duct PSC when no evidence for biliary tract disease was detected on cholangiography but histopathology from liver biopsy showed features of PSC and criterion 1 and 3 were present. Clinical data at diagnosis, laboratory findings, extension of biliary involvement (intrahepatic, extrahepatic, intrahepatic and extrahepatic, or small duct disease), and date of first biochemical sign or symptom of PSC were extracted from clinical records. The confluence of the bile ducts at the hilum was used to distinguish between the extra and intrahepatic bile duct system. In three cases, the diagnosis was established on symptoms, laboratory findings, and liver biopsy alone without any cholangiographic imaging performed; these cases were classified as unknown extension. Cases of PSC were classified as related or not related to IBD and large or small duct PSC. Overlapping autoimmune hepatitis was not considered a criterion for exclusion. Two different investigators, working in close contact, reviewed all cases. Cases with a doubtful diagnosis were discussed between all investigators in order to obtain a uniform case ascertainment.
Annual age-stratified data on the population of Västra Götaland was obtained from Statistics Sweden (www.scb.se). The national registration at Statistics Sweden was used to establish residency in all cases in order to identify patients who resided outside Västra Götaland at the time of diagnosis or moved outside the region during the study period and to obtain dates of death.
Point prevalence of PSC was calculated from the number of PSC patients aged ≥18 years residing in Västra Götaland as of December 31, 2005. Patients who were diagnosed outside Västra Götaland and moved into the region before this date were included in the analysis. Patients who had died, undergone liver transplantation, or moved outside the region before this date were excluded from prevalence estimates. Incident cases of PSC were defined as patients ≥18 years of age who fulfilled the criteria for PSC and resided in Västra Götaland at the time of diagnosis. Patients who were diagnosed outside Västra Götaland or resided outside the region at the time of diagnosis were excluded from incidence estimates. Age at diagnosis was categorized into nine groups: 18-25 years, 25-35 years, 35-45 years, 45-55 years, 55-65 years, 65-75 years, 75-85 years, 85-95 years, and >95 years. Changes in age and sex composition of the background population was adjusted for by using the annual age- and sex-specific population of Västra Götaland as a denominator in all incidence rate calculations. Crude incidence was calculated as the average annual incidence for the period 1992-2005. Poisson regression was used to simultaneously analyze temporal, age-associated, and sex-associated effects on the overall incidence rate of PSC and specific incidence rates of large duct PSC and PSC with or without IBD. Exact Poisson regression was used to analyze time trends in the incidence of small duct PSC due to the small number of cases in this category. Both age-adjusted and age-stratified trends were investigated. Average annual percentage change (AAPC) with a 95% confidence interval (CI) was obtained by multiplying incidence rate ratios related to year of diagnosis obtained in the adjusted Poisson model by 100. The effect of age on the incidence of PSC was investigated using age categories as a continuous variable. For all statistical analyses, a P value of less than 0.05 was considered statistically significant. SPSS version 15.0 software was used to calculate baseline characteristics. All regression analyses were performed using STATA 10 with robust standard error estimates.
Between 1992 and 2005, a total of 199 incident cases of PSC were detected in Västra Götaland; 142 (71.4%) were men and 57 (28.6%) were women. Diagnostic delay as estimated by time from first biological sign or clinical symptom of PSC until the establishment of the diagnosis could be assessed in 67% (134/199) of cases. The median diagnostic delay was 16 months; this time was unchanged when comparing the first and second half of the study period (16.1 versus 16.3 months, respectively). There was a change in the use of diagnostic tools during the study. Among patients diagnosed during the first half of the study period, 84% underwent endoscopic retrograde cholangiopancreatography (ERCP) and 46% underwent MRCP, whereas among patients diagnosed during the second half of the study period, 45% underwent ERCP and 77% underwent MRCP. The demographic and clinical characteristics of these patients are presented in Table 1.
Table 1. Incident Cases of PSC in an Adult Population of Västra Götaland, Sweden, 1992-2006
|Age in years, median (range)|| || ||38.5 (18.0-76.8)||38.7 (18.2-73.9)||37.6 (18.0-76.8)|
|Inflammatory bowel disease (IBD)||Yes||Any IBD||152 (76.4%)||117 (82.4%)||35 (61.4%)|
| || ||Ulcerative colitis||129 (64.8%)||104 (73.2%)||25 (43.9%)|
| || ||Crohn disease||17 (8.5%)||9 (6.3%)||8 (14.0%)|
| || ||Indeterminate colitis||5 (2.5%)||3 (2.1%)||2 (3.5 %)|
| || ||Microscopic colitis||1 (0.5%)||1 (0.7%)||0 (0%)|
| ||No|| ||47 (23.6%)||25 (17.6%)||22 (38.6%)|
|Extension of biliary involvement||Only intrahepatic|| ||61 (30.7%)||47 (33.1%)||14 (24.6%)|
| ||Extrahepatic and/or intrahepatic|| ||115 (57.8%)||81 (57.0%)||34 (59.6%)|
| ||Small duct disease|| ||20 (10.1%)||12 (8.5%)||8 (14.0%)|
| ||Unknown|| ||3 (1.5%)||2 (1.4%)||1 (1.8%)|
|Symptoms at diagnosis||Yes||Any symptom||93 (46.7%)||62 (43.7%)||31 (54.4%)|
| || ||Jaundice||48 (24.1%)||32 (22.5%)||16 (28.1%)|
| || ||Cholangitis||19 (9.5%)||13 (9.2%)||6 (10.5%)|
| || ||Itching||20 (10.1%)||14 (9.9%)||6 (10.5%)|
| || ||Abdominal pain||50 (25.1%)||34 (23.9%)||16 (28.1%)|
| || ||Other||66 (33.2%)||48 (33.8%)||18 (31.6%)|
| ||No|| ||106 (53.3%)||80 (56.3%)||26 (45.6%)|
Crude Incidence and Prevalence.
The mean crude annual incidence of PSC was 1.22 per 100,000 in the total population aged ≥18 years in Västra Götaland during the period 1992-2005. Among men and women, the incidence was 1.78 and 0.69, respectively, in the corresponding population. The point prevalence of PSC in the same population was 16.2 (23.7 among men and 8.9 among women) per 100,000 as of December 31, 2005.
Time Trends in Overall Incidence of PSC.
Trends in age-adjusted PSC incidence expressed as AAPC are presented in Table 2. The overall incidence of PSC increased significantly in the total adult population with an AAPC of 3.06 (95% CI 0.01-6.20) which equals an increase of 35.1 percent (95% CI 0.06-82.5) over a 10-year period. Sex-stratified analysis showed tendencies toward increasing trends in men (AAPC 2.30, 95% CI −1.35 to 6.08) and particularly in women (AAPC 4.98, 95% CI −0.46 to 10.72), but none of these tendencies were statistically significant.
Table 2. AAPC With 95% CI in the Incidence of PSC in Relation to Extension of Biliary Involvement and Presence or Absence of IBD
|All PSC||3.06 (0.01-6.20)||0.050||2.30 (−1.35 to 6.08)||0.220||4.98 (−0.46 to 10.72)||0.074|
|Extension|| || || || || || |
| Large duct||2.13 (−1.05 to 5.41)||0.192||0.63 (−2.95 to 4.34)||0.735||6.32 (0.03-13.02)||0.049|
| Small duct||8.31 (−3.17 to 21.76)||0.168||17.88 (0.95-40.25)||0.037||−3.11 (−19.22 to 15.67)||0.757|
|IBD|| || || || || || |
| No||5.91 (−0.64 to 12.89)||0.078||9.69 (0.82-19.33)||0.031||1.93 (−7.68 to 12.54)||0.705|
| Yes||2.22 (−1.25 to 5.82)||0.212||0.85 (−3.19 to 5.07)||0.684||7.01 (0.24-14.24)||0.042|
Time Trends in Incidence of Large and Small Duct PSC.
The time trends in age-adjusted incidence of large and small duct PSC are presented in Table 2. There was a significant increase in the incidence of large duct PSC among women (AAPC 6.32, 95% CI 0.03-13.02) but not among men (Table 2). The incidence of small duct PSC increased significantly among men (AAPC 17.88, 95% CI 0.95-40.25) but not among women (Table 2).
Time Trends in Incidence of IBD-Associated PSC.
Diverging trends were observed for the incidence of PSC related to IBD when comparing men and women (Table 2). In women, there was a significant increase in the incidence of IBD-associated PSC (AAPC 7.01, 95% CI 0.24-14.24), but no significant trends in the incidence of PSC without concurrent IBD (Table 2). In men, the incidence of PSC without IBD increased significantly (AAPC 9.69, 95% CI 0.82-19.33), but there was no statistically significant trend for IBD-associated PSC (Table 2).
Effects of Age on Incidence of PSC.
The age-related effects on the incidence of PSC are given in Table 3. Age had a strong and statistically significant influence on the incidence of IBD-associated PSC and large duct PSC, with a decreasing incidence with increasing age. There was no significant effect of age on the incidence of small duct PSC and PSC without IBD. These findings were consistent in both sexes.
Table 3. Incidence Rate Ratios for PSC Related to Age Stratifying for Extension of Biliary Involvement and Presence or Absence of IBD
|All PSC||0.81 (0.76-0.87)||<0.001||0.81 (0.75-0.89)||<0.001||0.81 (0.71-0.92)||0.001|
|Extension|| || || || || || |
| Large duct||0.82 (0.76-0.86)||<0.001||0.82 (0.75-0.89)||<0.001||0.84 (0.73-0.96)||0.009|
| Small duct||0.76 (0.57-0.98)||0.036||0.81 (0.55-1.13)||0.232||0.69 (0.42-1.05)||0.090|
|IBD|| || || || || || |
| No||1.01 (0.89-1.14)||0.872||1.05 (0.88-1.26)||0.588||0.97 (0.82-1.15)||0.731|
| Yes||0.75 (0.69-0.82)||<0.001||0.77 (0.69-0.85)||<0.001||0.70 (0.58-0.85)||<0.001|
We investigated the prevalence and incidence of PSC in a defined adult population in a regional health care unit in Sweden. The overall incidence rate of PSC increased significantly over the investigated period. Stratified analysis revealed significantly increasing trends for IBD-associated PSC and large duct PSC among women and for PSC without IBD and small duct PSC among men. The prevalence of PSC at the end of the study period (16.2/100,000) is the highest reported to date. To our knowledge, this is the largest population-based study of the incidence and prevalence of PSC. The study's strengths are the population-based methodology, the size of the cohort, and the standardized validation of all cases.
This is the first population-based study showing a statistically significant increase in the incidence of PSC. Trends in the overall incidence of PSC have been investigated previously in two studes from the United States 11 and the United Kingdom. 14 Both of these studies reported a tendency toward increasing incidence that did not reach statistical significance, possibly due to a lack of power owing to the relatively small cohorts. We also found significant trends in different subtypes of PSC with an increase of small duct PSC and PSC not associated with IBD in men and large duct PSC and IBD-associated PSC in women. Although these trends were statistically significant, the possibility of a type I error (i.e., that these trends appeared by chance) has to be considered, as should the relatively small number of cases in each subgroup. Studies on the total incidence of IBD in Sweden during the investigation period are lacking. One study, which investigated the incidence of Crohn's disease in Stockholm, reported increasing incidence from 1991 to 1996 and thereafter a decrease in incidence from 1996 to 2001. 15 A study from Olmsted County, Minnesota, reported increasing incidence of IBD until the early 1970s, after which the incidence rates for both Crohn disease and ulcerative colitis stabilized. 16 In our study, there was a significant increase in the incidence of IBD-associated PSC but not in PSC without IBD in men, and the opposite situation was observed in women, with a significantly increasing incidence of PSC without IBD but not with IBD. The reasons for these diverging trends are unclear.
We found a crude incidence of 1.22 and a prevalence of 16.2 per 100,000 in the present study. This finding indicates a disease duration of about 13 years, which is in accordance with the literature. 4-7 The incidence and prevalence of PSC have been investigated previously in four small population-based studies (each including 17-53 cases) from Norway, 10 the United States, 11 Wales, 12 and Canada. 13 Incidence rates of 0.9-1.3 and prevalence rates of 8.5-13.6 per 100,000 person-years have been reported in these studies. Limited data exist on the incidence of PSC in Asia; one publication from Singapore indicates a significantly lower incidence than in western countries. 17 In a recent study from the United Kingdom, the incidence and prevalence of PSC was studied in the General Practice Research Database. 14 Even though that study was not based on a general population, it is less likely to encounter the problems of studies performed at referral centers with biased selection of cases and difficulties in determinating the population at risk when calculating incidence ratios. It was also considerably larger than previous population-based studies, including 149 incident cases with PSC. Markedly lower prevalence (3.85/100,000) and incidence (0.41/100,000) was reported in this study. Because these figures represent the incidence and prevalence in the General Practice Research Database and hence not the general population, they are not directly comparable with our results. Even lower incidence and prevalence rates have been reported from Spain (0.07/100,000 and 0.08/100,000, respectively), 18 the authors speculate that this may be explained by the low prevalence of IBD in Spain. A detection bias through underreporting may also have influenced these results, because the study relied on active reporting of cases by the treating physician through a questionnaire.
The male/female ratio in our series (71% male patients) is similar to what has been reported in population-based studies from Norway (71%) and the United States (68%) but is slightly higher than what has been observed in the United Kingdom (62%-64%) and Canada (55%). 10-14 The diagnosis of PSC was associated with IBD in 76% of all cases in our study, a finding that is similar to estimates from the all other population-based studies except for Card et al. 14 from the United Kingdom, who reported a significantly lower prevalence of IBD in their population (48%); however, the investigators in that study did not have access to patient files, so some IBD diagnoses may have been missed. The median age at presentation in the present study was 38.5 years, which is also highly similar to all of the aforementioned studies, with the exception of the United Kindgdom studies, in which the mean and median age at presentation was around 15 years higher. 12, 14 When comparing data on age at presentation between these studies, it is essential to bear in mind that we excluded pediatric cases from our series, whereas Kaplan et al. 13 included them. The remaining population-based studies do not state whether a search for PSC patients was conducted at pediatric clinic registries. 10, 11, 14
Not surprisingly, the association between age and the overall incidence of PSC was statistically significant. However, stratified analysis revealed that this association was only present in IBD-associated PSC; the incidence of PSC without IBD was not associated with age. This aspect of PSC has not been studied, and further investigation into the clinical characteristics and natural course of PSC with and without concurrent IBD is warranted.
Certain possible methodological limitations of the study have to be taken into consideration. As in most retrospective studies, a detection bias could produce false temporal trends. The International Classification of Diseases (ninth and tenth revision) codes are not exclusive for PSC but comprise all forms of cholangitis, and careful validation of the PSC diagnosis is therefore mandatory. In the present study, we used the Mayo criteria for the validation of the PSC diagnosis. 11 Follow-up data in case records were scrutinized for other potential etiologies of cholangitis, and cases were excluded if there was a suspicion of secondary cholangitis. Applying these strict criteria for the diagnosis, we believe that the risk for overestimation of prevalence and incidence rates due to a detection bias is very low. We have only been able to search diagnostic registers from departments of surgery and internal medicine for diagnostic codes for PSC. Consequently, cases that never have been seen by a surgeon or an internist/gastroenterologist have not been identified. However, patients with PSC in Sweden are routinely managed by gastroenterologists and in some cases surgeons, whereas general practitioners do not provide care for these patients. Some cases may have been missed, because the early stages of PSC may be without any symptoms. However, because diagnostic registers were searched until the end of 2006, we believe that most patients with an uncertain diagnosis at the end of the study period would have presented a more clear clinical picture during the following year given the natural course of the disease. New diagnostic tools or increased availability of established methods can introduce another form of detection bias. The use of MRCP has increased during the study period and has partly replaced ERCP as the standard diagnostic tool (presumably because of the risk of ERCP-related complications). A possible detection bias related to this shift from ERCP to MRCP has to be considered even though the result of such a bias is difficult to predict. The sensitivity of MRCP for PSC changes in the biliary tree is probably inferior to that of ERCP, 19 which can be predicted to result in a false decrease in the incidence of PSC. However, this false trend could be compensated by a bias in the opposite direction if the total number of investigated patients and/or investigations per patient increased due to the less invasive nature of MRCP. Considering the progressive nature of PSC, one would expect that sooner or later the diagnosis will be made regardless of the imaging method used. Any significant detection bias would therefore manifest itself as a change in the diagnostic delay (i.e., the time from the first biochemical sign or symptom of the disease until the time when the diagnosis is established). In our study, the diagnostic delay was constant over the entire study period. This implies that although a detection bias related to the shift from MRCP to ERCP cannot be excluded completely, it is unlikely that it is important enough to have had any significant impact on the observed trends. The increasing use of MRCP may also have led to a classification bias, because the sensitivity for subtle changes in the large bile ducts may be different from previous diagnostic methods, leading to an illusory change in the ratio between the incidence of large and small duct PSC. We did not observe any trends in the incidence of small and large duct PSC that would raise such a suspicion. On the contrary, trends in small duct PSC were opposite in men and women and a detection bias in this aspect therefore seems unlikely since the same diagnostic work up was used in both sexes.
In conclusion, we investigated the incidence and prevalence of PSC in a defined population in southwestern Sweden. We confirm the clinical characteristics of PSC reported in smaller population-based studies. In addition, we provide new and reliable estimates of the prevalence and incidence of PSC, and our study is the first to report a significantly increasing incidence of PSC. This finding may be due to improved ascertainment of the diagnosis; however, it may also reflect a true increase of PSC, suggesting that the medical burden of this disease may be higher than previously estimated.