Primary Biliary Cirrhosis Once Rare, Now Common in the United Kingdom?

Transcription

1 Primary Biliary Cirrhosis Once Rare, Now Common in the United Kingdom? OLIVER F. W. J AMES, 1 RAJ BHOPAL, 2 DENISE HOWEL, 2 JACKIE GRAY, 2 ALASTAIR D. BURT, 1 AND JANE V. METCALF 1 There is a widespread impression that the number of patients with the autoimmune liver disease primary biliary cirrhosis (PBC) is increasing, although its incidence and prevalence vary widely. Using thorough case-finding methods and rigorous definitions to assess changes in incidence and prevalence with time and to explore the symptomatology and mortality of the disease in a large group of unselected patients, we performed a descriptive epidemiological study of PBC in a well defined population over a fixed period of time using established diagnostic criteria and with clinical follow-up of all cases. In a population of 2.05 million in northern England 770 definite or probable PBC cases were identified. Prevalence rose from per 10 6 in the adult population and per 10 6 women over 40 in 1987 to per 10 6 adults and per 10 6 women over 40 in Incidence was 23 per 10 6 in 1987 and 32.2 per 10 6 in Three hundred patients died in median follow-up of 6.27 years (141 liver deaths); the standardized mortality ratio was At presumed diagnosis, 60.9% had no symptoms of liver disease. By June % of prevalent patients had liver symptoms. PBC is apparently increasing. It is still unclear whether this is because of a true increase, case finding, or increased disease awareness. The study draws attention to (1) high mortality from liver disease and non liver-related causes even in patients initially with no liver symptoms and (2) apparently poor diagnostic awareness of the disease. (HEPATOLOGY 1999;30: ) Recently interest in the autoimmune liver disease primary biliary cirrhosis (PBC) has increased considerably. 1,2 This may be attributed not only to advances in understanding of the immune mechanisms of the disease but also to increasing recognition of apparent variation in its incidence and prevalence. The tools of epidemiology may offer understanding of the true clinical spectrum of the disease as well as offering clues as to its etiology. 3 Most available data on the epidemiology of PBC has hitherto been drawn from case series. 4 The Abbreviations: PBC, primary biliary cirrhosis; AMA, antimitochondrial antibodies; LFT, liver function test; ONS, Office of National Statistics. From the Centre for Liver Research, 1 and the Department of Epidemiology and Public Health 2 University of Newcastle upon Tyne, Newcastle upon Tyne, UK. Received March 22, 1999; accepted May 4, J.V.M. was a Wellcome Trust Training Fellow. Address reprint requests to: Professor Oliver F.W. James, Centre for Liver Research, University of Newcastle, The Medical School, Framlington Place, Newcastle upon Tyne, NE2 4HH, England. O.F.W.James@ncl.ac.uk; fax: (44) Copyright 1999 by the American Association for the Study of Liver Diseases /99/ $3.00/0 390 few studies using case finding methods in defined populations over specific time periods have reported relatively small numbers of patients. 5-9 We recently suggested guidelines for the correct conduct of a descriptive epidemiological study of PBC to obtain the best available data from such a study and to approach as closely as possible information concerning the true incidence and prevalence of the disease. 4 These guidelines are (1) stringent case inclusion criteria; (2) definition of date of disease onset; (3) well-defined study period, area, and population; (4) multiple case finding methods; and (5) rigorous tracing of all possible cases. The objects of the present study were to use these guidelines to perform a comprehensive case finding study to describe the epidemiology of PBC in a population of sufficient size in which (1) sufficient numbers of cases could be described to provide meaningful overall figures on incidence, prevalence, and mortality and (2) an examination of a mode of presentation or detection of the patients could be made, thus providing some estimate of possible extent and reasons for under-reporting in earlier case finding studies. PATIENTS AND METHODS Case Definition Definite PBC. Three criteria had to be fulfilled for definite PBC: antimitochondrial antibody (AMA) positive (indirect immunofluorescence 1 in 40), abnormal liver function tests (LFTs; at least one of serum bilirubin, serum alkaline phosphatase, and serum alanine transaminase), and diagnostic or compatible liver histology (Scheuer) (all traceable biopsy specimens examined independently by A.D.B.). Probable PBC. PBC was probable if 2 of the previous criteria were fulfilled. If no liver histology, then AMA was positive and LFTs were abnormal. If AMA was negative or less than 1 in 40, then liver histology was diagnostic of PBC and LFTs were abnormal. Patients initially classified as probable (fulfilling 2 criteria) who subsequently fulfilled the third (usually by having a liver biopsy) were reclassified as definite. The results reported are based on final case status within the study period (up to December 31, 1994). Date of Diagnosis The date of diagnosis was defined as the earliest date at which the patient was found (often by retrospective examination of clinical case records) to have fulfilled any 2 of the 3 diagnostic criteria. This was to avoid the need for different criteria for date of diagnosis in the asymptomatic versus the symptomatic group of patients. This was called the date of presumed diagnosis. Study Population The area of study comprised 6 adjacent health areas of North-East England including Northumberland, Sunderland, North Durham, South Durham, South of Tyneside, and North of Tyneside. Enquiry from regional hospital activity analysis has shown almost no flow of patients from this region to hospitals outside the region. Boundaries

2 HEPATOLOGY Vol. 30, No. 2, 1999 JAMES ET AL. 391 within the region were verified from the Office of National Statistics (ONS), postcode (zipcode), and local health authority data. Midyear population estimates for each year were based upon United Kingdom 1991 Census data and ONS projections. Study Period and Calculation of Rates The period over which the case finding exercise took place was January 1, 1987, to December 31, Incidence rates per million population at risk were calculated for the period of January 1 to December 31 each year. Point prevalence rates were obtained (per million) for June 15 each year (the Census date). Full follow-up on mortality data was obtained until December 31, Rates for adults aged 20 years and over were included to aid comparison with previous studies and for women aged 40 years and over because PBC is a disease affecting predominantly this section of the population. Case Finding Methods The case finding methods have been previously described. 8 They were briefly as follows: (1) Requests were made to all physicians in internal medicine within the region to identify all cases of possible PBC under their care. (2) A search was done of hospital admission data on Regional Information Systems for all 13 hospitals in the region for all admission episodes using ICD-9 code 571.6, which includes all cases of biliary cirrhosis (both primary and secondary). The new ICD-10 coding came into use in April 1994, thus code K74-3 (PBC) was used from this date to the end of the study period. (3) Examination was done of all hospital immunology data for patients with positive AMA reported in 1 in 40 by indirect immunofluorescence M2 antibody confirmatory testing in 1 laboratory for 1 year only (1994). Initially there was only one central immunology laboratory, but by the end of the study period there were 8 laboratories. About 207,000 out of approximately 253,000 (82%) autoimmune profiles processed by the laboratories over the 8-year period were available for examination. (4) All listings from the ONS of deaths within the region and in the study period in which ICD-9 code or (subsequently) ICD-10 K74.3 appeared anywhere on the death certificate were examined. Data Collection Hospital Case Records. Hospital case records were exhaustively sought in all cases except in the very few individuals who had never attended a hospital but who were detected via the autoantibody screen (requested by a primary health care physician) or death certificate (with no hospital records known). Other Cases. Cases who were no longer under continuing care of a hospital consultant or who had never attended a hospital were traced via their physician or the Family Health Service Authority. The physician was contacted and (1) asked for clinical details and (2) asked for permission for J.V.M. and O.F.W.J. to write to the patients with a request to attend the PBC clinic so that possible PBC could be evaluated. Permission was also sought from hospital consultants to allow all individuals still attending a hospital for reasons unconnected with possible PBC, and who were not already known to have PBC, to be offered an outpatient appointment at the PBC Clinic at Freeman Hospital, Newcastle upon Tyne. Permission was also sought from the physicians of these patients before this offer. They were assessed clinically and via laboratory test profile. If they had not had a liver biopsy this was offered if clinically indicated. All cases were flagged to the ONS so that all deaths (and copies of death certificates) were notified to the study group. Confirmation of survival of all subjects not notified in this way was obtained at the end of the study period either from the PBC clinic records, other hospital records, or from the patients physician. Symptom Status Symptoms defining cases of symptomatic PBC were persistent right upper quadrant abdominal pain, persistent ( 3 months) fatigue, pruritus, clinically apparent jaundice, ascites, bleeding varices, and encephalopathy. Patients recorded as having any of these symptoms were regarded as symptomatic. Patients without the above were regarded as asymptomatic of PBC. Deaths Death certificates were obtained from the ONS, and clinical case records of all patients notified dead from January 1, 1987, to December 31, 1996, were examined and the most likely cause of death was determined from the case notes. Cause of death was attributed as liver related or as non liver related. Database Design A relational database was designed a Microsoft Office Access Database (Microsoft). Data were entered onto the database from data collection sheets filled in from each set of patient clinical hospital records. All matters referring to collection, handling, storage, and access to this data conformed to the Data Protection Act. Statistical Analysis The pattern of prevalence and incidence rates over the years was analyzed using a 2 test for trend. Person-years at risk were calculated for each patient starting at presumed diagnosis of PBC and ending with death or end of follow-up period. Annual mortality rates for all causes by gender and 5-year age bands in England and Wales were obtained for the years 1987 to 1996 from ONS. The standardized mortality ratios and their 95% confidence intervals based on the Poisson distribution were then calculated. Ethical Approval Ethical Approval was obtained from Ethical Committees of all health districts in the study area. RESULTS Cases and Case Definitions A total of 1,140 individuals were identified for possible inclusion in the study. Of these, 770 were identified who fulfilled at least 2 case criteria (definite or probable PBC) by December 31, 1994, and who lived within the study area and were alive at some time between January 1, 1987, and December 31, Of the 770 patients included in the study, 472 fulfilled all 3 diagnostic criteria and were considered as definite cases, whereas the remaining 298 cases fulfilled only 2 criteria and were described as probable (Table 1). Reasons for exclusion of the other cases were as follows: residency beyond study area (77), death prior to January 1, 1987 (94), and diagnosis after December 31, 1994 (75). Other reasons for exclusion included physicians refusing permission to approach patients, patients refusing to attend the clinic, if the patient died before other investigations (except AMA) could be performed (20), or if repeat AMA testing negative 2 or more times with no other evidence to support diagnosis of Category TABLE 1. Diagnostic Criteria for Potential PBC No. of Cases Abnormal LFTs Positive AMA Compatible Liver Biopsy Definite 472 Probable (298) 263 ND 12 Normal 23 Negative NOTE. Total definite and probable cases 770 (included). Abbreviation: ND, not done.

3 392 JAMES ET AL. HEPATOLOGY August 1999 PBC (14, including 5 children). Final exclusion criteria included incomplete data on residence (3), incompatible liver biopsy (15), positive AMA but normal LFTs, or no liver biopsy (72). Of the 770 cases 76 were men; female: male ratio was 9.1: 1. Case Identification and Case Finding Among the 770 definite or probable cases, 296 (38.4%) diagnosed as a result of investigations for symptoms and/or signs of liver disease were being carried out. In a further 264 patients (34.3%), the diagnosis of presumed PBC (date at which any 2 of the 3 diagnostic criteria were fulfilled and recorded in clinical case records) was made retrospectively by the investigators from careful review of the clinical case records among patients in whom investigation of nonhepatic symptoms and/or of other diseases was being carried out. In 30 patients it was unclear from the clinical case records why investigations had been performed (earliest available case records showed diagnosis already made, so reasons for original investigations not clear). A total of 180 patients (23.4%) had the diagnosis made after identification and, if necessary, further investigation by case finding exercises (134 by this study, 46 during previous much less exhaustive studies in 1979 and 1984). Review of all the case records suggested that PBC was diagnosed or suspected by the original hospital consultant under whom investigations had been performed from which a diagnosis of definite or probable PBC could be made in 441 (57.3%) patients. The diagnosis was neither made nor apparently considered in 286 patients (37%). In the majority of these (222) only laboratory investigations (positive AMA and abnormal LFTs) were recorded in the case records. In 66 cases there were also typical symptoms (42), signs (5), or both (19) recorded in the case records, yet the diagnosis had still not been apparently considered. It was unclear whether the diagnosis had been considered in 43 patients (5.7%). Population Studies The total population studied (based on the ONS midyear population estimates corrected for the 1991 census) varied between 2,034,760 and 2,052,668 during the years 1987 to The population of patients over age 20 was between 1,505,792 and 1,538,642. The female population of patients over age 40 was between 493,124 and 502,469. Incidence There were 468 new incident cases during the study period (423 women, 45 men; Table 2). Of these, 239 were definite cases, 13 were AMA negative autoimmune cholangiopathy cases, 6 had normal LFTs but positive AMA and diagnostic biopsy, and 210 had abnormal LFTs and positive AMA but had not undergone liver biopsy. Incidence varied between 47 (23 per 10 6 ) and 66 (32.2 per 10 6 ) cases per year. Among adults (over age 20) this was 31.2 per 10 6 to 43.0 per 10 6 cases per year. Among women over 40 it was 91.3 per 10 6 to 116 per 10 6 per year. The mean annual incidence from 1987 to 1990 was 26.1 per 10 6, and the mean annual incidence from 1991 to 1994 was 31.1 per Formal tests for trend show that this pattern is not statistically significant for definite and probable cases in all subjects, those over age 20, or females over age 40 (P.075, P.089, and P.23, respectively). Year Prevalence TABLE 2. Mean Annual Incidence and Prevalence for Total and Definite Cases by Years Whole Population Incidence Adults H20 Yr Women H40 Yr D P D D P D D P D Prevalence Abbreviations: D, definite; P, probable. On June 15, 1987 there were 304 prevalent cases (275 women of whom 263 were over age 40) and 29 men (overall prevalence 149 per 10 6 ; Table 2). On June 15, 1994 prevalence had risen to 514 cases (473 women, 467 over age 40) and 41 men (overall prevalence per 10 6 ). Prevalence among the adult population in 1987 was per 10 6 and among the female population over age 40 in 1987 was per By 1994 these figures had risen to per 10 6 (adult population) and per 10 6 (women over 40). The increase in prevalence of definite and probable PBC is highly significant (P.00001). Symptom Status and Mortality Symptoms. Among the 590 patients in whom a presumed diagnosis of PBC was made (either by the original physicians or retrospectively by the present investigators) either during investigation of possible liver disease or incidentally during investigation of other nonhepatic diseases or nonhepatic symptoms, 262 (44.4%) had symptoms of PBC at the time of presumed diagnosis. (Note definitions of date of presumed diagnosis was the date at which the patient was found on review of case records to have fulfilled any 2 of the 3 diagnostic criteria). Of the remaining 180 cases in whom the diagnosis was made after case finding by the present investigators, 39 (21.7%) of these had symptoms of PBC or bilirubin greater than 50 mmol/l (clinical jaundice) recorded in their case records at the time of presumed diagnosis on our review after retrieval. Thus, at presumed diagnosis, 301 of 770 (39.1%) patients had recorded symptoms of PBC, and 469 (60.9%) were asymptomatic or had no recorded symptoms. In June 1994, 514 cases were prevalent, and at the most recent clinical assessment (by individual follow-up) 62% of these had experienced PBC symptoms and 38% had never experienced symptoms and were, therefore, still asymptomatic.

4 HEPATOLOGY Vol. 30, No. 2, 1999 JAMES ET AL. 393 Mortality. Of the 770 patients (median follow-up 6.27 years) 300 had died by December 31, There were 171 liver deaths or transplants (33 transplants, 3 of whom died), and therefore, the total liver deaths were 141. Among the 469 initially asymptomatic patients, 69 (14.7%) died liver deaths (57) or were transplanted (14, 2 died), and of the 301 initially symptomatic patients, 102 (33.8%) died liver deaths (84) or were transplanted (19, 1 died). There was a total of 4,331 years of follow-up between 1987 and 1996, (mean 5.62 years) during which there were 300 deaths from any cause and 159 deaths from causes other than liver disease. The standardized mortality ratio was % confidence interval (2.54 to 3.19) indicating death rates from all causes in PBC patients, which is nearly 3 times as high compared with the general population after adjusting for age and sex. The standardized mortality ratio for all causes excluding liver disease was % confidence interval (1.41 to 1.91). DISCUSSION This is an extensive study of the epidemiology of PBC. It continues and very greatly extends an earlier study (published in 1990, containing information up to 1987) concerning part of the same region. 7 We have recently published information concerning the incidence and prevalence of PBC in the city of Newcastle upon Tyne with a population of 280,000 within the population of 2.05 million in the present study. A total of 160 of the present 770 patients were included in this earlier study. 8 We have recently critically reviewed information concerning earlier studies of PBC epidemiology and demography 4 but will not now repeat this exercise. Nonetheless, it is important to state that the present study has used a widely accepted valid case definition of PBC in a rigorously defined population over a well defined period of time. We suggest that it is valid to use the figures for definite and probable patients rather than the definite patients alone in the main presentation of results for the following reasons. Among the probable patients it seems very likely that the vast majority do have PBC. Indeed 37 of these died of liver disease and 87 had typical symptoms of PBC with the other possible causes of liver disease excluded. It is important to emphasize that 263 of 298 of these patients are classified as probable because no liver histology was obtained and they therefore were only able to fulfill 2 of the 3 diagnostic criteria. It was deemed that liver biopsy was not ethically justifiable in many of these patients who were elderly and/or asymptomatic of liver disease. A further proportion of patients refused liver biopsy; in addition, a small number of patients with clinically advanced cirrhosis and poor coagulation did not undergo biopsy. If there is an error of inclusion, patients included who did not have PBC, this must be small. Thus, only 15 of 487 (3.1%) patients who had abnormal LFTs and positive AMA and who had liver biopsy were found to have liver histology incompatible with PBC; that is to say they had no features compatible with or diagnostic of PBC and had features diagnostic of another liver pathology and were hence deemed not to have PBC (and were excluded from the study). The potential error of excluding all of the 263 nonbiopsied patients from the calculations of prevalence and incidence would, we suggest, be far greater. In respect of the 12 patients with positive AMA and diagnostic/compatible histology but normal LFTs, we have recently shown that at least 80% of such patients will ultimately develop abnormal LFTs and in over 75% symptoms of liver disease. 10 It could indeed be argued that we should also have included the 72 patients with positive AMA but normal LFTs in whom no liver biopsy was performed because many of these will ultimately develop classical PBC. 10 We have, however, not done so, preferring to adhere to our case definitions. In respect to the 23 patients who were AMA negative, but had abnormal LFTs and diagnostic liver histology, we feel that we are justified to include them because AMA-negative PBC is widely regarded as part of the spectrum of PBC Because of our case finding methods we may have underestimated the prevalence and proportion of AMA-negative PBC in our population. Taken in conjunction with an earlier much less comprehensive study in our region, the present study shows a very striking increase in the prevalence of PBC in the region from 16 per 10 6 (definite cases only) in 1976 to per 10 6 in This prevalence figure has until very recently been considerably higher than in any other study. It is of interest, therefore, that a new study from Wales in which no case finding methods were used (only a careful case register) has suggested a comparable prevalence figure of 200 per 10 6 in a population of 250,000 in In our region incidence also showed a trend toward increase, rising from a mean 11.3 per 10 6 in 1976 to 1980 and 18.8 per 10 6 in 1983 to to 26.1 per 10 6 in 1987 to 1990 and 31.1 per 10 6 in 1991 to 1994, although there was considerable year-on-year variation. There are 4 possible explanations for the very striking increase in prevalence, none of which are mutually exclusive: (1) There is a true increase in cases of the disease; it is becoming much more common. (2) There is an increasing recognition of the disease as our case finding methods have become more exhaustive. (3) Patients with PBC are living longer, hence more were prevalent in 1994 (because of better prognosis) than 10 or 15 years earlier. (4) Earlier diagnosis is resulting in longer survival due to lagtime bias. At present we are not able to offer definite data to distinguish these explanations. The rate of increase of the prevalence of the disease appears consistent since Efforts at case finding have taken place within the region since 1979, and one of us (O.J.) has had an acknowledged interest in PBC since that time. Nonetheless, more physicians with interest in gastroenterology have been appointed in the 13 hospitals in the region (1976, n 3; 1987, n 10; 1994, n 15). A total of 354 of the 590 patients not found by case finding (60%) had attended Freeman Hospital Liver Clinic. In support of the idea that increased recognition of the disease plays an important part in the apparent increased prevalence it should be noted that in the years before 1976, PBC was thought of as a cholestatic disease, usually presenting with clinical jaundice. Among the present 770 patients only, 51 (6.6%) had bilirubin over 50 mmol/l (3 mg%) at presumed diagnosis (i.e., they were overtly jaundiced); these would presumably correspond to the majority of patients diagnosed in the early 1970s. Furthermore, to explain the increase in prevalent cases from 304 on June 15, 1987 to 514 on June 15, 1994, only on the basis of a true increase in the disease, a rise in incidence of 26 cases per year (total 210 in 8 years) would be implied. This is not borne out by the incidence data, although some increase was seen. It is of interest that no significant rise in incidence of definite PBC was seen over the study period, whereas a more marked rise in probable PBC was seen, possibly reflecting our case-finding efforts and increased diagnostic activity (particularly use of autoanti-

5 394 JAMES ET AL. HEPATOLOGY August 1999 body screens) over the study period. On the other hand, we believe it is unlikely that the increase in prevalent cases can be entirely explained either by increased interest in PBC/ diagnostic activity or by improvements in treatment. Liver transplantation was only performed on 33 patients and, although about 100 were receiving ursodeoxycholic acid treatment by 1994, it is unlikely that this will have contributed very substantially to increased prevalence via a very marked increase in longevity. The extent to which longer survival due to lagtime bias may contribute to the apparent increase in prevalence will need further long-term analysis of mortality data. The data from the case finding part of the study is relevant to earlier studies from elsewhere. It should be noted that almost a quarter (180 or 23.4%) of the patients were detected during the case-finding exercise. Furthermore, although on review of case records 21.7% of these 180 had recorded symptoms that could be attributed to PBC or bilirubin greater than 50 mmol/l (3 mg%) there was no sign whatever that the diagnosis had been previously considered or mentioned. Indeed, in the total patient population, review of the case records suggests that at a time when a presumed diagnosis of PBC could be made it had not been considered by the responsible consultant in 286 (37.1%) patients of whom 66 had abnormal LFTs, were AMA positive, and had recorded symptoms and/or signs of liver disease. Thus, at best, surveys relying on known cases of PBC may underestimate prevalence by over one quarter and more probably by over one third It is of considerable interest that at the last census date (June 1994), just over three fifths of the prevalent (62%) patients had experienced symptoms attributable to PBC, so just under two fifths (38%) had yet to develop symptoms. This probably gives a best estimate of what proportion of patients are symptomatic in a community at any one point in time. Long follow-up of the cohort of 469 patients who were initially asymptomatic of liver disease at the time of presumed diagnosis will eventually give us the best idea of symptom development and the natural history of the disease (almost no asymptomatic patients have received ursodeoxycholic acid treatment). Previous studies of the development of symptoms in initially asymptomatic patients have been drawn from case series from major tertiary referral centers, thus being subject to selection bias, the largest of which examined 95 patients We have shown significantly increased death rates in PBC patients compared with the general population. This has been observed before, but the adjustment for age and sex was performed by a method which does not give an estimate of the magnitude of the difference and these were much smaller case series. 22,23 However, we have been able to estimate that the death rates for PBC patients in this region are approximately three times those for the general population (if all causes are considered), and one and a half times as high if deaths from liver disease are excluded (the cause for this new observation is, as yet, unclear). In summary, we have shown an extraordinary increase in prevalence of PBC to the highest figure in the world, with less increase in incidence. The present data do not allow us to make definite conclusions as to whether the increase is because of a true increase of the disease and/or to increased interest in PBC, diagnostic activity, and attendant factors. We have shown that the PBC was widely underdiagnosed; this may be relevant to other studies. Finally, the data emerging from this cohort support earlier contentions as to the importance of PBC in relation to mortality even in initially asymptomatic patients. 20,21 REFERENCES 1. Gershwin ME, MacKay IR. Primary biliary cirrhosis part 1. Semin Liver Dis 1997;17: Kaplan MM. Primary biliary cirrhosis. N Engl J Med 1996;335: Metcalf JV, Howel D, James OFW, Bhopal RS. Primary biliary cirrhosis epidemiology helping the clinician. Br Med J 1996;312: Metcalf JV, James OFW. The geoepidemiology of primary biliary cirrhosis. Semin Liver Dis 1997;17: Danielsson A, Boqvist L, Uddenfeldt P. Epidemiology of primary biliary cirrhosis in a defined rural population in the northern part of Sweden. HEPATOLOGY 1990;11: Lofgren J, Jarnerot G, Danielsson D, Hemdal I. Incidence and prevalence of primary biliary cirrhosis in a defined population in Sweden. Scand J Gastroenterol 1985;20: Myszor M, James OFW. The epidemiology of primary biliary cirrhosis in northern England: an increasingly common disease? QJM 1990;75: Metcalf JV, Bhopal RS, Gray J, Howel D, James OFW. Incidence and prevalence of primary biliary cirrhosis in the city of Newcastle upon Tyne, England. Int J Epidemiol 1997;26: Kingham JGC, Parker DR. The association between primary biliary cirrhosis and coeliac disease: a study of relative prevalences. Gut 1998;42: Metcalf JV, Mitchison HC, Palmer JM, Jones DE, Bassendine MF, James OFW. Natural history of early primary biliary cirrhosis. Lancet 1996;348: Michieletti P, Wanless IR, Katz A, Scheuer PJ, Yeaman SJ, Bassendine MF, Palmer JM, et al. Antimitochondrial antibody negative primary biliary cirrhosis: a distinction syndrome of autoimmune cholangitis. Gut 1994;35: Lacerda MA, Ludwig J, Dickson ER, Jorgensen RA, Lindor KD. Antimitochondrial antibody-negative primary biliary cirrhosis. Am J Gastroenterol 1995;90: Invernizzi P, Crosignani A, Battezzati PM, Covini G, De Valle G, Larghi A, Zuin M, et al. Comparison of the clinical features and clinical course of antimitochondrial antibody-positive and -negative primary biliary cirrhosis. HEPATOLOGY 1997;25: Almdal TP, Sorensen TI. Incidence of parenchymal liver diseases in Denmark, 1981 to 1985: analysis of hospitalization registry data. The Danish Association for the Study of the Liver. HEPATOLOGY 1991;13: Remmel T, Temmel H, Uibo R, Salupere V. Primary biliary cirrhosis in Estonia, with special reference to incidence, prevalence, clinical features, and outcome. Scand J Gastroenterol 1995;30: Watson R, Goss B, Angus P. Low prevalence of primary biliary cirrhosis (PBC) in Victoria, Australia. Gut 1995;36: Witt-Sullivan H, Heathcote J, Cauch K, Blendis L, Ghent C, Katz A, Milner R, et al. The demography of primary biliary cirrhosis in Ontario, Canada. HEPATOLOGY 1990;12: Boberg KM, Aadland E, Jahnsen J, Raknerud N, Stiris M, Bell H. Incidence and prevalence of primary biliary cirrhosis, primary sclerosing cholangitis and autoimmune hepatitis in a Norwegian population. Scand J Gastroenterol 1998;33: Mitchison HC, Lucey MR, Kelly PJ, Neuberger JM, Williams R, James OFW. Symptom development and prognosis in primary cirrhosis: a study in two centres. Gastroenterology 1990;99: Balasubramanian K, Grambsch PM, Wiesner RH, Lindor KD, Dickson ER. Diminished survival in asymptomatic primary biliary cirrhosis: a prospective study. Gastroenterology 1990;90: Mahl RC, Shockcor W, Boyer JL. Primary biliary cirrhosis: survival of a large cohort of symptomatic and asymptomatic patients followed for 24 years. J Hepatol 1994;20: Dickson ER, Grambsch PM, Fleming TR, Fisher LD. Prognosis of primary biliary cirrhosis: model for decision making. HEPATOLOGY 1989;10: Christensen E, Altman D, Neuberger J, De Stavola BL, Tygstrup N, Williams R. Updating prognosis in primary biliary cirrhosis using a time-dependent Cox regression model. Gastroenterology 1993;105: