The Rate of Decompensation and Clinical Progression of Disease in People with Cirrhosis
K. M. Fleming; G. P. Aithal; T. R. Card; J. West
Authors and Disclosures
Posted: 03/01/2011; Alimentary Pharmacology & Therapeutics. 2010;32(11):1343-1350. © 2010 Blackwell Publishing
Abstract
Background
We lack population-based estimates of the rate of decompensation in people with compensated cirrhosis as well as estimates of the manner in which the disease progresses once identified.
Aim
To determine the rate of decompensation and clinical progression of disease in patients with cirrhosis based upon clinical symptoms recorded electronically in general practice data.
Methods Using Cox proportional hazards regression, we modelled the rate of decompensation for patients from the UK General Practice Research Database with a diagnosis of cirrhosis between 1987 and 2002. We determined the clinical progression in the first year following diagnosis and subsequently categorizing patients through time according to a simple clinical staging system agreed at the Baveno IV consensus conference.
Methods Using Cox proportional hazards regression, we modelled the rate of decompensation for patients from the UK General Practice Research Database with a diagnosis of cirrhosis between 1987 and 2002. We determined the clinical progression in the first year following diagnosis and subsequently categorizing patients through time according to a simple clinical staging system agreed at the Baveno IV consensus conference.
.
Results
Results
The rate of decompensation in patients with compensated cirrhosis was found to be 11% overall. The rate of decompensation was higher in the first year (at 31% compared with 7.3% afterwards) and in patients with an alcoholic aetiology. Patients with compensated cirrhosis had a 1-year probability of proceeding directly to death of 7% compared with 20% in patients with decompensated cirrhosis.
.
Conclusions
Conclusions
Using data recorded in general practice records, it is possible to determine the rate of decompensation and the clinical progression of disease in people with cirrhosis.
Introduction
The natural history of cirrhosis has long been of interest to gastroenterologists and hepatologists[1–7] as knowledge of the progression of the disease allows planning of management and development of effective strategies for the primary prevention of complications and is of obvious interest to patients who wish to understand their risk of death, distressing symptoms and time to referral for consideration for transplantation. While in recent years advances have been made in the management of portal hypertension[8] with a consequent apparent improvement in some outcomes,[9] we have surprisingly little information on the effect of these developments on the natural history of cirrhosis in a general population-based setting. The most commonly referenced studies have focused on those patients who are admitted to hospital. In addition, while scoring systems such as Child-Pugh, MELD, UKELD enable the ranking of patients at an individual level and near the end stage of disease, we cannot at present provide useful absolute estimates of the rate of decompensation, clinical progression or mortality to patients in the earlier stages of disease.
In a recent comprehensive systematic review of the natural history and prognostic indicators of survival in cirrhosis which included 118 pertinent publications,[10] the authors reported a wide range of estimates of survival and factors that predicted death, but many methodological difficulties were also described. Perhaps most saliently they provided for the first time 1-year outcome probabilities for cirrhosis according to the clinical stages of disease agreed at the Baveno IV consensus conference.[11] However, as their estimates were based on the experience in a sole hospital in Sicily in the late 1970s–early 1980s, they may not be widely generalizable. We therefore do not have contemporary estimates of the progression of disease in cirrhosis at a population level.
Increasingly people are being diagnosed with cirrhosis at earlier stages of the disease i.e. compensated rather than decompensated. We have consequently taken the opportunity presented by prospectively recorded electronic primary care data from the UK to determine the rate of decompensation and the progression of disease in patients with cirrhosis based upon the Baveno IV stage of disease criteria.
Description of GPRD
The General Practice Research Database (GPRD) is a longitudinal database consisting of anonymous computerized general practice records for over 13 million patients in the UK, including over 50 million patient-years of usable data. Practices are subjected to regular data quality checks and audits to maintain 95% inclusion of prescribing and morbidity events. Data contained within this database are recorded through direct entry during general practice appointments and following communication from secondary care. Data are coded based on both the Oxmis and Read medical coding dictionaries. The GPRD has previously been shown to be broadly representative of the population of the UK[12] of which it represents approximately a 4–6% sample. Approval for this study was given by the Scientific and Ethical Committee of the GPRD.
Study Population
We obtained all records of patients with a diagnostic code for any liver disease within the GPRD between June 1987 and April 2002. Patients aged 25 years or over were then selected based on the presence of a diagnostic or therapeutic code for cirrhosis, oesophageal varices and/or portal hypertension to represent a cohort of adult diagnosed cirrhosis. This age cut-off was designed to avoid including patients who may have had the onset of cirrhosis under the age of 18. Each patient was assigned a date of diagnosis of cirrhosis as the date of the first record of any of these codes.
Definitions
We then categorized patients with cirrhosis through time following date of diagnosis according to the four stages of cirrhosis as agreed at the Baveno IV conference[11]–
Introduction
The natural history of cirrhosis has long been of interest to gastroenterologists and hepatologists[1–7] as knowledge of the progression of the disease allows planning of management and development of effective strategies for the primary prevention of complications and is of obvious interest to patients who wish to understand their risk of death, distressing symptoms and time to referral for consideration for transplantation. While in recent years advances have been made in the management of portal hypertension[8] with a consequent apparent improvement in some outcomes,[9] we have surprisingly little information on the effect of these developments on the natural history of cirrhosis in a general population-based setting. The most commonly referenced studies have focused on those patients who are admitted to hospital. In addition, while scoring systems such as Child-Pugh, MELD, UKELD enable the ranking of patients at an individual level and near the end stage of disease, we cannot at present provide useful absolute estimates of the rate of decompensation, clinical progression or mortality to patients in the earlier stages of disease.
In a recent comprehensive systematic review of the natural history and prognostic indicators of survival in cirrhosis which included 118 pertinent publications,[10] the authors reported a wide range of estimates of survival and factors that predicted death, but many methodological difficulties were also described. Perhaps most saliently they provided for the first time 1-year outcome probabilities for cirrhosis according to the clinical stages of disease agreed at the Baveno IV consensus conference.[11] However, as their estimates were based on the experience in a sole hospital in Sicily in the late 1970s–early 1980s, they may not be widely generalizable. We therefore do not have contemporary estimates of the progression of disease in cirrhosis at a population level.
Increasingly people are being diagnosed with cirrhosis at earlier stages of the disease i.e. compensated rather than decompensated. We have consequently taken the opportunity presented by prospectively recorded electronic primary care data from the UK to determine the rate of decompensation and the progression of disease in patients with cirrhosis based upon the Baveno IV stage of disease criteria.
Description of GPRD
The General Practice Research Database (GPRD) is a longitudinal database consisting of anonymous computerized general practice records for over 13 million patients in the UK, including over 50 million patient-years of usable data. Practices are subjected to regular data quality checks and audits to maintain 95% inclusion of prescribing and morbidity events. Data contained within this database are recorded through direct entry during general practice appointments and following communication from secondary care. Data are coded based on both the Oxmis and Read medical coding dictionaries. The GPRD has previously been shown to be broadly representative of the population of the UK[12] of which it represents approximately a 4–6% sample. Approval for this study was given by the Scientific and Ethical Committee of the GPRD.
Study Population
We obtained all records of patients with a diagnostic code for any liver disease within the GPRD between June 1987 and April 2002. Patients aged 25 years or over were then selected based on the presence of a diagnostic or therapeutic code for cirrhosis, oesophageal varices and/or portal hypertension to represent a cohort of adult diagnosed cirrhosis. This age cut-off was designed to avoid including patients who may have had the onset of cirrhosis under the age of 18. Each patient was assigned a date of diagnosis of cirrhosis as the date of the first record of any of these codes.
Definitions
We then categorized patients with cirrhosis through time following date of diagnosis according to the four stages of cirrhosis as agreed at the Baveno IV conference[11]–
.
stage 1, uncomplicated cirrhosis i.e. cirrhosis in the absence of both oesophageal varices and ascites;
stage 2, cirrhosis with oesophageal varices, but without ascites and without bleeding;
stage 3, cirrhosis with ascites, with or without oesophageal varices;
and stage 4, cirrhosis with GI bleeding, with or without ascites.
.
Stages 1 and 2 together correspond to compensated cirrhosis with stages 3 and 4 together corresponding to decompensated cirrhosis. Each patient was assigned up to four dates corresponding to the earliest recording of the symptoms representing each stage. To minimize the potential for misclassification of patients with decompensated cirrhosis as patients with compensated disease we additionally looked for the recording of one or more prescriptions for spironolactone as evidence of decompensation.
Death was defined using a combination of the patient's registration status within the GPRD and medical codes for death, with the earliest date of these being assigned as the date of death.
Presumed aetiology of cirrhosis was defined as alcoholic, viral hepatitis, autoimmune liver disease or metabolic liver disease as previously reported.[13] For the purposes of analyses, all cases not defined as alcoholic were classified as non-alcohol-related cirrhosis.
Statistical Analyses
Using Cox proportional hazards, we modelled the rate of decompensation in the group of patients with compensated disease. Subjects entered the analysis period at the date of earliest recorded code for stage 1 or 2 and exited at the earliest of date of earliest recorded code for stage 3 or 4, date of death, date of liver transplant, date of deregistration with their general practice or 30 April 2002, which was the last date of available data in this dataset. We examined the rate of decompensation for subjects with alcoholic cirrhosis and those with non-alcohol-related cirrhosis. Follow-up time was split at 1 year and we modelled the rate of decompensation during the first year after diagnosis and subsequent to that first year again stratifying by presumed aetiology.
We examined the probabilities of patients with cirrhosis progressing from an individual stage to a subsequent stage or death within 1 year and across the whole time period of their records with 95% confidence intervals being calculated assuming a binomial distribution. We again stratified the cohort into those with alcoholic cirrhosis and those with non-alcohol-related cirrhosis.
Results
A total of 4537 patients were identified with cirrhosis. Table 1 shows selected clinical and demographic information about the study population: 56.1% entered the study cohort at stage 1, 12.8% at stage 2, 23.0% at stage 3 and 8.1% at stage 4. A total of 2307 (50.9%) of cirrhosis cases were classified as alcoholic cirrhosis with the remainder being assigned as non-alcohol-related cirrhosis for the purposes of the subsequent analyses.
Rate of Decompensation
Overall, the rate of decompensation for those patients with compensated disease was 11.8% per year adjusted for age and gender [95%CI (11.2%, 12.5%)]. This figure varied considerably both by length of follow-up and by presumed aetiology. A Kaplan–Meier survival curve showing the difference in rate of decompensation by presumed aetiology is shown in Figure 1.
Death was defined using a combination of the patient's registration status within the GPRD and medical codes for death, with the earliest date of these being assigned as the date of death.
Presumed aetiology of cirrhosis was defined as alcoholic, viral hepatitis, autoimmune liver disease or metabolic liver disease as previously reported.[13] For the purposes of analyses, all cases not defined as alcoholic were classified as non-alcohol-related cirrhosis.
Statistical Analyses
Using Cox proportional hazards, we modelled the rate of decompensation in the group of patients with compensated disease. Subjects entered the analysis period at the date of earliest recorded code for stage 1 or 2 and exited at the earliest of date of earliest recorded code for stage 3 or 4, date of death, date of liver transplant, date of deregistration with their general practice or 30 April 2002, which was the last date of available data in this dataset. We examined the rate of decompensation for subjects with alcoholic cirrhosis and those with non-alcohol-related cirrhosis. Follow-up time was split at 1 year and we modelled the rate of decompensation during the first year after diagnosis and subsequent to that first year again stratifying by presumed aetiology.
We examined the probabilities of patients with cirrhosis progressing from an individual stage to a subsequent stage or death within 1 year and across the whole time period of their records with 95% confidence intervals being calculated assuming a binomial distribution. We again stratified the cohort into those with alcoholic cirrhosis and those with non-alcohol-related cirrhosis.
Results
A total of 4537 patients were identified with cirrhosis. Table 1 shows selected clinical and demographic information about the study population: 56.1% entered the study cohort at stage 1, 12.8% at stage 2, 23.0% at stage 3 and 8.1% at stage 4. A total of 2307 (50.9%) of cirrhosis cases were classified as alcoholic cirrhosis with the remainder being assigned as non-alcohol-related cirrhosis for the purposes of the subsequent analyses.
Rate of Decompensation
Overall, the rate of decompensation for those patients with compensated disease was 11.8% per year adjusted for age and gender [95%CI (11.2%, 12.5%)]. This figure varied considerably both by length of follow-up and by presumed aetiology. A Kaplan–Meier survival curve showing the difference in rate of decompensation by presumed aetiology is shown in Figure 1.
Figure 1.
Kaplan–Meier survival estimates for decompensation in subjects with compensated alcoholic and non-alcohol-related cirrhosis.
During the first year after diagnosis, the rate of decompensation was 31.0% adjusted for age and gender [95%CI (28.8%, 33.4%)]; for those patients with a presumed aetiology of alcoholic cirrhosis, the rate of decompensation during the first year was 37.6% [95%CI (34.1%, 41.5%)] compared with 25.2% [95%CI (22.6%, 28.2%)] for those with non-alcohol-related cirrhosis.
Following the first year, the rate of decompensation did not vary much by presumed aetiology with a rate of 7.3% per year adjusted for age and gender [95%CI (6.5%, 8.2%)] for those with alcoholic cirrhosis compared with an adjusted rate of decompensation of 5.5% per year [95% CI (4.8%, 6.2%)] for those with non-alcohol-related cirrhosis.
Progression
Outcome probabilities in the first year are shown in Figure 2. Patients in stages 1 and 2 had a 1-year probability of proceeding directly to death of around 7%. The probability of progression to death within 1 year was highest for patients within stage 3, rather than stage 4 (at 20.1% and 18.2% respectively), although this difference was not statistically significant.
Figure 2.
Probabilities for progression within 1 year, all patients.
State progressions across the whole study period are shown in Figure 3 Of note is the high proportion of people in the early stages of disease i.e. patients with compensated cirrhosis, who progressed directly to death without the recording of other clinical symptoms of decompensation. These results support the categorization of stages 1 and 2 together as representing compensated disease and stages 3 and 4 together as representing decompensated disease.
Figure 3.
Probabilities for progression across whole record, all patients.
The probabilities of progressing both during the first year and state progressions subsequently for those with alcoholic cirrhosis and non-alcohol-related cirrhosis are shown in Table 2 and Table 3 respectively. Reading across the table shows the stage at which patients began their follow-up. The rows in the column represent the stage to which the patients directly progressed. For example, looking at the first column, we can see that alcoholics who began follow-up in stage 1 had a 64.6% chance of remaining in stage 1 during the first year, with 2.9% progressing directly to stage 2 (oesophageal varices), and so on. In the first year following diagnosis, patients with alcoholic cirrhosis had a slightly higher probability of progressing to another stage of cirrhosis than those with non-alcohol-related cirrhosis, but a lower probability of progressing directly to death. Across the whole study period, patients with a presumed aetiology of alcoholic cirrhosis were again more likely to progress to another later stage of cirrhosis than those with non-alcohol-related cirrhosis. The differences in progression directly to death were not so marked across the whole time period, rather less patients with alcoholic cirrhosis remained in their entry stage with the exception of patients in stage 3.
Discussion
Using the clinical observations recorded in primary care records of people with cirrhosis, we have described the progression of disease as experienced by patients and as recorded by primary care physicians in the UK. Progression of disease is rapid with the rate of decompensation in patients with compensated cirrhosis being 11% per year, but is particularly rapid in the first year following diagnosis at 31% in this first year. We have used a relatively simple staging system, based on the presence or absence of symptoms rather than relying on laboratory tests, mimicking the Baveno IV consensus clinical staging system to classify patients with cirrhosis. The classification is useful in long-term planning of management of patients with cirrhosis. We have applied this classification to an ambulatory population from primary care to describe the natural history of this disease. For example, in those patients without recorded evidence of portal hypertension, i.e. those in stage 1 of disease, there was about a 11% probability of progressing to a more severe form of disease (development of ascites, varices or GI bleeding) within a year alongside a 7% likelihood of dying. In those patients with stage 2 disease, the chance of developing ascites or bleeding was approximately 24% in the following year alongside a 7% risk of death.
By virtue of the electronic primary care data from a broad sample of primary care physicians at our disposal, we have constructed a large, representative, population-based cohort of patients with cirrhosis. This cohort was identified reasonably recently (1987–2002) and therefore the results we have generated reflect the natural history of cirrhosis during this period. As this is a population-based cohort, it is unlikely to have been affected by the variation in referrals and follow-ups seen in cohorts selected from secondary care. We are confident based upon our own previous validation study (which showed the vast majority of patients with a recorded code for cirrhosis had available extra evidence from secondary care[13] that for the diagnosis of cirrhosis in general this coding within primary care is good. The additional clinical signs and symptoms of decompensation may not always be recorded as accurately, however, unless they are of obvious clinical relevance to the GP. Hence, ascites is probably symptomatic and hence likely to be moderate to large-volume, clinically significant ascites (rather than that only identified by ultrasound). We acknowledge that there may have been some misclassification between stages 1 and 2 as the recognition of development of oesophageal varices in the absence of bleeding depends on endoscopic surveillance which, for the period our data cover, was not necessarily uniform in the UK. We believe though that it is unlikely that much misclassification has occurred between the compensated and decompensated disease states, nor between stages 3 and 4 as events such as bleeding oesophageal varices are of such importance that they are highly unlikely to be inaccurately recorded.
For the purposes of predicting progression of disease and mortality, we were unable to calculate the typically used Child-Pugh classification[14,15] or the more recently derived model for end-stage liver disease (MELD)[16] as we had no measures of blood indices available. However, we argue that the strength of Baveno IV consensus clinical staging system is that it allows such a broad evaluation of the natural history of patients with cirrhosis without relying upon laboratory tests. In addition, unlike Child-Pugh and MELD scores which are of limited value in early stages of cirrhosis where specific laboratory tests are unaltered (e.g. Child-Pugh class A patients), the Baveno classification can be applied to both early stage disease and late stage disease. This not only allows us to describe disease progression, but also to do so in a way which is intelligible both to nonspecialist clinicians and to patients. Of course, the simplicity of the Baveno classification is also one of its weaknesses in that many other factors that are known to influence the progression of cirrhosis, such as the development of hepatocellular carcinoma, hepatic encephalopathy and nutritional status are not included within the criteria.
We have included both incident and prevalent cases of cirrhosis by design as there is clear evidence that cirrhosis, as with most chronic diseases, has a wide heterogeneity of presentation. The results of our study therefore reflect the real world of clinical practice within the general population, which is unlikely to be true of series collected from within secondary care centres with a particular interest in the disease. This may explain some of the discrepancies with previously reported data as discussed below.
In comparison with historical studies of the natural history of cirrhosis,[1–6,10] our work shows that even in the later stages of the 20th century, progression of disease is rapid. Following our replication of outcome model of D'Amico et al. [10] using the Baveno IV stage of disease system, we found both similarities and differences with their reported findings. We found a similar clear demarcation of risk of death between compensated disease (stage 1 and stage 2) and decompensated disease (stage 3 and 4), although the probability of death within 1 year we observed for stage 1 disease and stage 2 disease was substantially higher than they reported. In contrast, we did not find a difference in the rate of progression to death for those patients with ascites rather than bleeding oesophageal varices. The explanation for these disparities most likely lies in the differences in methodology and population selection we have used. In the current study, we have taken each individual's record and defined his/her progression of disease within 1 year, whereas D'Amico et al. appear to report outcome probability estimates, which were averaged risks from 10 years of follow-up. In addition, we have used a population-based approach to collection of cases rather than the single centre approach previously used. The great benefits of a population-based approach i.e. avoiding bias and improving generalizability need no rehearsal, but in this case, the potential for misclassification of minor degrees of decompensation consequent upon lack of direct access to cases may explain some of the difference in our results. It is likewise probable that much of the difference between our results is a consequence of the fact that the UK, with low levels of chronic viral hepatitis, has a different aetiological picture with respect to cirrhosis from that seen in Sicily (the setting from which the Baveno classification was derived). Finally, some of the difference may be due to the era of data collection. Our finding that ascites is as severe a marker of poor prognosis as bleeding fits with the prevailing view that since the late 1970s, widely available important improvements have been made in the management and outcomes of variceal bleeding[17] with improvements in the treatments for ascites broadly limited to the administration of antibiotics in cases of spontaneous bacterial peritonitis, which is a much less frequent complication than bleeding.
A recent paper from Denmark looking at only alcoholic cirrhosis has provided similarly greater estimates for the 1-year progression directly to death in patients with no reported complications of 10% and similar estimates for 1-year progression directly to death for patients with clinical symptoms of ascites and variceal bleeding.[18] There are no such comparable figures for patients with non-alcohol-related cirrhosis available in current published literature.
In summary, our study has described the disease progression following a diagnosis of cirrhosis in contemporary clinical practice using data recorded electronically in general practice. Our reproduction of the use of a simple, internationally recognized, Baveno IV consensus clinical staging system is practical, useful and relevant to patients with cirrhosis and clinicians in the UK and beyond (within the limits of the assumptions underpinning its design). Of particular interest is the overall rate of decompensation for all those with cirrhosis as a contemporary estimate of this has not been readily available. Clearly, further validation of the approach we have used to stage disease in other datasets will be of value.
We conclude that our results in conjunction with the previously observed increasing incidence of cirrhosis emphasize the growing threat of liver disease to public health.
Abstract and Introduction
Methods
Results
DiscussionAbstract and Introduction
Methods
Results
References
No comments:
Post a Comment