Acute Liver Failure -- One Disease, More Than 40 Definitions
Acute Liver Failure -- One Disease, More Than 40 Definitions
In this review we identified, categorised and listed definitions of ALF used in evaluation studies. Large variations in the definitions of ALF were found and worse, some studies did not report any explicit definition of disease. It is conceivable that some retrospective observational studies introduced modifications to original definitions to allow statistical analysis with the existing data; in case not all necessary data were available.
To our knowledge this is the first systematic review exclusively dedicated to variability of definitions of this disease. We allowed all synonyms of ALF in our search and we found our conclusions on a large number of studies. A limitation of the search, however, is that we only addressed studies in which the prognostic effect of indicators formed a main objective; we may consequently have missed studies with a more limited focus on prediction. The strength of our study is its extensiveness based on a large number of studies.
The limitation of our study could be a bias by extracting definitions from the studies. We extracted the definitions based on the full text reported in the studies and in case when no full text of the definition is reported, but it is only in the reference, we use the definition of the reference. Sometimes a definition was explicitly given in the text and a reference for that definition was also provided. Nevertheless we used full text definition. However, in some cases the explicit definition was not consistent with the reference. Some examples are: the definition includes presence of prothrombin activity of less than 40%, whereas the authors refer to Trey and Davidson where PT was not included; similarly, a definition reported occurrence of HE within 8 weeks, but the authors reported using the definition of O'Grady, which corresponds to 12 weeks. Undoubtedly this way of reporting brings confusion. We would like to underline the importance of exact citing of originally proposed definitions. When a study proposes its own definition this should be reported explicitly.
The most often used definition, 19 times, was the one proposed by Trey and Davidson. It is important to clarify that one should not think this definition is better than others, based on the number of studies using this definition. This is partly explained by the fact that it was the first definition of ALF in the literature. As shown in Table 3 this definition was not used in the last few years and one may expect that other 'younger' definitions will earn more attention in the coming years.
Among the studies reporting the occurrence of HE within some time period following the onset of symptoms or onset of disease, most of them did not define 'symptoms'and/or 'onset of disease'. Fourteen studies clearly specified time interval as the time between HE and onset of jaundice. Notably using different terms, for onset of disease brings confusion and clear specification is required. It would be less confusing if one generally accepted term was used, e.g. 'jaundice', defined by a threshold of plasma bilirubin.
Another critical component of ALF definitions is the degree of coagulopathy, but only 42 studies reported the presence of coagulopathy as an element of the definition. We believe that assessment of the level of coagulopathy is necessary to diagnose the severity of ALF. According to most of the studies coagulopathy is operationalised by PT or INR. Note that there are different thresholds for PT and INR values. A comparison of studies becomes even more difficult when in some studies PT is expressed as percentage of normal value and in others as prolongation (in seconds). The values for PT depend on the used method and normal values. However, there is rather large variation in laboratory assays, mainly due to the source of the used thromboplastin. For this reason INR seems to be more appropriate. Unfortunately, to date there is no uniform standardisation of measuring coagulopathy in ALF patients.
Table 1 presents an overview of components used in ALF definitions and may serve as a tool to create a uniform definition. In our opinion a proposal for a draft definition of ALF should be based on the following: absence of underlying liver disease of any sort, presence of HE of any grade, occurring within 8 weeks after onset of jaundice (defined as plasma bilirubin >50 μm) together with presence of coagulopathy, defined by INR (≥1.5).
In sum, the ambiguity and variability in the definitions of ALF hamper the comparability among the studies. However, the exact implications of the differences in ALF definitions on performance of prognostic markers are unknown yet. Although meta-analysis is theoretically possible, due to heterogeneity the results will not be meaningful. There is room for improvement in the reporting of ALF definitions in future prognostic studies. The result of this review may be useful as a starting point to create a uniform ALF definition.
Discussion and Conclusion
In this review we identified, categorised and listed definitions of ALF used in evaluation studies. Large variations in the definitions of ALF were found and worse, some studies did not report any explicit definition of disease. It is conceivable that some retrospective observational studies introduced modifications to original definitions to allow statistical analysis with the existing data; in case not all necessary data were available.
To our knowledge this is the first systematic review exclusively dedicated to variability of definitions of this disease. We allowed all synonyms of ALF in our search and we found our conclusions on a large number of studies. A limitation of the search, however, is that we only addressed studies in which the prognostic effect of indicators formed a main objective; we may consequently have missed studies with a more limited focus on prediction. The strength of our study is its extensiveness based on a large number of studies.
The limitation of our study could be a bias by extracting definitions from the studies. We extracted the definitions based on the full text reported in the studies and in case when no full text of the definition is reported, but it is only in the reference, we use the definition of the reference. Sometimes a definition was explicitly given in the text and a reference for that definition was also provided. Nevertheless we used full text definition. However, in some cases the explicit definition was not consistent with the reference. Some examples are: the definition includes presence of prothrombin activity of less than 40%, whereas the authors refer to Trey and Davidson where PT was not included; similarly, a definition reported occurrence of HE within 8 weeks, but the authors reported using the definition of O'Grady, which corresponds to 12 weeks. Undoubtedly this way of reporting brings confusion. We would like to underline the importance of exact citing of originally proposed definitions. When a study proposes its own definition this should be reported explicitly.
The most often used definition, 19 times, was the one proposed by Trey and Davidson. It is important to clarify that one should not think this definition is better than others, based on the number of studies using this definition. This is partly explained by the fact that it was the first definition of ALF in the literature. As shown in Table 3 this definition was not used in the last few years and one may expect that other 'younger' definitions will earn more attention in the coming years.
Among the studies reporting the occurrence of HE within some time period following the onset of symptoms or onset of disease, most of them did not define 'symptoms'and/or 'onset of disease'. Fourteen studies clearly specified time interval as the time between HE and onset of jaundice. Notably using different terms, for onset of disease brings confusion and clear specification is required. It would be less confusing if one generally accepted term was used, e.g. 'jaundice', defined by a threshold of plasma bilirubin.
Another critical component of ALF definitions is the degree of coagulopathy, but only 42 studies reported the presence of coagulopathy as an element of the definition. We believe that assessment of the level of coagulopathy is necessary to diagnose the severity of ALF. According to most of the studies coagulopathy is operationalised by PT or INR. Note that there are different thresholds for PT and INR values. A comparison of studies becomes even more difficult when in some studies PT is expressed as percentage of normal value and in others as prolongation (in seconds). The values for PT depend on the used method and normal values. However, there is rather large variation in laboratory assays, mainly due to the source of the used thromboplastin. For this reason INR seems to be more appropriate. Unfortunately, to date there is no uniform standardisation of measuring coagulopathy in ALF patients.
Table 1 presents an overview of components used in ALF definitions and may serve as a tool to create a uniform definition. In our opinion a proposal for a draft definition of ALF should be based on the following: absence of underlying liver disease of any sort, presence of HE of any grade, occurring within 8 weeks after onset of jaundice (defined as plasma bilirubin >50 μm) together with presence of coagulopathy, defined by INR (≥1.5).
In sum, the ambiguity and variability in the definitions of ALF hamper the comparability among the studies. However, the exact implications of the differences in ALF definitions on performance of prognostic markers are unknown yet. Although meta-analysis is theoretically possible, due to heterogeneity the results will not be meaningful. There is room for improvement in the reporting of ALF definitions in future prognostic studies. The result of this review may be useful as a starting point to create a uniform ALF definition.
Source...