Summary of main findings
The BALLETS study is the first GP based study where the entire cohort was comprehensively tested for additional diseases (such as viral hepatitis) after an abnormal LFT, using the full analyte panel and normal reference ranges. We have shown that an abnormal LFT alone does not select out a population where the prevalence rate approaches a threshold which would justify viral screening. We have assessed the validity of the various strategies a GP could adopt, at least as far as viral hepatitis is concerned, when faced with an abnormal LFT of uncertain provenance. The intuitive response for a GP in such a situation would be to repeat the LFT, an approach advocated by current literature. This study shows that this may not be the optimal policy. This strategy is the most expensive, even more so than viral testing all patients, as the costs incurred include repeating the LFT as well as viral testing the majority. The study also shows that if ALT is notably raised (greater than twice the upper limit of normal), then the probability of chronic viral hepatitis is high (nearly 8%), but sensitivity is low. The strategy of testing all people from prevalent countries is the second most efficient, in terms of cost per case detected, and detects almost twice as many cases as the most efficient strategy - testing for viral infection when two conditions (birth in a prevalent country and an ALT greater than twice the upper limit of normal) are satisfied. The relative financial disadvantages of the strategy of repeating the LFT would be even greater if patient costs were included, as the extra visit would have to be factored in.
Previous literature on LFTs and liver diseases
We conducted a literature review using the search strategy shown in Table , with the aim of retrieving papers that studied a cohort of patients with abnormal LFT results to provide evidence on the probability of various liver diseases (including chronic viral hepatitis) given
abnormal test results. Any such studies would enable the precision of our observations to be strengthened. The search strategy returned 1,448 papers, including a previous review by Dufour et al. (2000) [29
Search strategy for studies looking prospectively at patients who have received an abnormal LFT*
Only eight studies matched our requirement of following up patients with an abnormal LFT result. Two additional articles were selected from the references of relevant studies. As a result, to the best of our knowledge, there are only ten studies where a cohort of asymptomatic patients with abnormal LFTs were followed up (Table ). However, one article was written in Korean (only the abstract was translated), so was excluded from our analysis.
Studies that have followed up patients from the general population after abnormal LFT
Two of the remaining nine English language papers described record linkage studies. One such study was based on the Korean insurance database that was linked with death certificates [30
]. This study reported that increased ALT, even within the upper end of the normal range, was associated with eventual death from liver disease. A study carried out in Scotland linked general practice and hospital databases [31
]. However, this was a retrospective study so a full liver screen was not conducted and follow-up was for a median of four years only, whereas many diseases, including chronic viral hepatitis, have much longer prodromal periods [33
The other seven studies were prospective cohort studies, based on testing asymptomatic members of the general population. The famous Dionysos study based on three analytes from the LFT analyte panel [34
] is included among these. In this study an impressive 6,917 citizens from two communities in northern Italy were screened. Although they tested all those who had an abnormal LFT (n = 1473) for viral hepatitis, for which they found a prevalence rate of 2.4%, they did not describe the pattern of LFT results in infected patients. Another Italian study by Pendino and colleagues (2005) screened 1,645 inhabitants from a town in southern Italy, with both a LFT (ALT, AST and GGT) and viral screen [35
]. The prevalence of viral hepatitis is much higher in this region because of a significant immigrant population, and they performed a more extensive analysis on the impact of viral hepatitis on LFTs. Of the 319 (19.4%) individuals who received an abnormal LFT, nearly 18% were infected with viral hepatitis. However, the LFT missed 34 (40%) of the 92 cases of viral hepatitis present in the community. Perhaps the most comprehensive prospective analysis looking at the effect of viral hepatitis on the individual analytes was carried out on a population of Japanese office workers [16
]. The study used data from compulsory health checks, which included an ALT, AST and GGT panel along with certain additional tests, such as a viral screen, that were added for study purposes. They found that ALT was the most sensitive of the three analytes used, detecting nearly half the cases of viral hepatitis, whilst being abnormal in 14% of the cohort (278 abnormal results in 1,973 participants). The remaining four prospectively designed studies were carried out in general practices and were therefore closer in population terms to the BALLETS cohort. However, three of these were restricted to patients with persistently abnormal LFTs over a six month period [36
] and one of these did not include a test for viral hepatitis. The final prospective study by Whitehead (1999) was small and based on only one analyte [39
After this review of the literature we concluded that there has been no published study that fully investigated a cohort of patients in primary care with an abnormal LFT result (from the full LFT analyte panel).
Strengths and limitations of the study
The main strength lies in the unique nature of the BALLETS cohort, being the only prospective study that has looked at the consequences of an abnormal LFT from a full analyte panel in primary care. The main limitation of our study relates to the rather small number of cases of chronic viral hepatitis (n = 13) and hence wide confidence limits on the results. That said the results are plausible, in the sense that they are consistent with the pathophysiology of hepatitis and in line with what was found in non-practice settings (see literature review above). They are available for meta-analysis with potential future studies.
We deliberately selected multi-cultural inner city populations in order to provide a sizable sub-group of people from countries where chronic viral hepatitis is common, as a result of infection during infancy (hepatitis B),[40
] and iatrogenic infection (hepatitis C).
Our study considers only one disease type, chronic viral hepatitis, while GP decision making must take into account other diseases, such as haemochromatosis, as well as other behavioural and social motivations for testing [41
]. That said, our conclusion that repeating the LFT "offers more than it delivers," may well apply to diseases such as PBC and haemochromatosis.
Lastly we have presented an analysis for cost minimisation and incremental cost per case detected. This is not a full cost-effectiveness or decision analysis. Donnan et al. did attempt a decision analysis [32
]. However this decision analysis was intended to find the most cost-effective strategy in the short term and used a limited time horizon of one year. LFTs are often ordered to prevent poor outcome in the long term, with many serious liver diseases, viral hepatitis included, manifesting over decades. Anxiety resulting from a false positive result was included in the model while long term health gains as a result of successful case finding and treatment were not captured.
Neither our decision analysis, nor that in Donnan's HTA report [32
], considered cost-effectiveness. We tackle this limitation by considering our results in the context of published cost-effectiveness analyses for screening for viral hepatitis (i.e. studies that found screening was cost-effective in populations with high prevalence rates e.g. migrants) and attempt to produce a "fast and frugal heuristic" [28
] to guide practice.
Implications for practice: a fast & frugal heuristic
The intuitively appealing practice of repeating abnormal LFTs (strategy A) gets little support from our analysis. It is the most expensive option, both in absolute terms and in terms of cost per case detected, compared to all five alternative strategies (Table ) - including that of simply testing everyone for viral infection.
The most important question a doctor can ask a patient with abnormal LFTs is their country of origin. This holds good whether the person settles in an area of high or low ethnic mix, since infections are acquired in infancy (hepatitis B), or as a result of sub-standard medical practices, such as needle sharing (hepatitis C). Once infected, people "take their risk with them" - less people will need to be tested in a low ethnic mix area, but those from prevalent countries still need testing. The strategy of testing people from prevalent countries promises good value for money. In this study, 11 of the 13 cases originated in medium or high risk countries. Thus the prevalence of chronic hepatitis viral infection (positive predictive value) among people with an abnormal LFT who were born in a prevalent country was 6.5% (11/170, 95% CI: 3.7%-11.2%, see table ), while the prevalence among the home born population (of all ethnic groups) was less than 0.2% (2/1038, CI: 0.05%-0.7%). Our findings support viral testing only in the former group, consistent with the threshold prevalence for both HBV and HCV, of approximately 3% at which population screening becomes cost-effective [21
Four of the strategies, C, D, E and F, entail viral testing in a population where the rate of hepatitis exceeds the 3% threshold for which testing has proven cost-effective in screening programs (Table ). The cost-effective threshold is probably a little lower in a diagnostic population than in a screening population (costs of inviting people to attend are lower and cases detected might be a slightly higher risk) but no other strategy yields a population with a hepatitis rate exceeding even 2%.
Strategy D (test immigrants from prevalent countries) has a better (lower) incremental cost-effective ratio than C and detects twice as many cases as E. However, the strategy F, testing immigrants from prevalent countries or any people with a very high ALT, is our preferred strategy, being both sensitive and efficient. We therefore recommend the "fast and frugal" heuristic described in Figure . This combines strategy F with normal judgement of clinical indications. For example a patient who is an intravenous drug user, or who has recently returned from a trip abroad where they had an attack of hepatitis, would be tested notwithstanding the result of the LFTs. Otherwise we recommend testing all patients with an abnormal LFT who were born in a country of intermediate or high prevalence and all patients for whom the ALT exceeds twice the limit of normal.
Fast and frugal heuristic decision tree.
The probability of chronic viral hepatitis is low even when the ALT exceeds this limit and the patient does not originate from a medium or high-risk country (1.6%). Nevertheless we advocate testing in these patients for the following reasons:
1. It is hard to ignore a level this high, and the wide confidence levels from our data suggest the need for flexibility [45
2. The progression for undetected chronic viral hepatitis is worse for patients with ALT greater than twice the upper limit of normal, and this level has been used as a threshold for treatment in guidelines.
3. If chronic viral hepatitis is not present at this level a more in-depth search for other causes of hepato-cellular damage is indicated.
We draw the line on further viral testing after this algorithm has been followed, unless of course further clinical indicators emerge. The likelihood of a case of viral hepatitis being present following the exclusions in this algorithm is approximately 0.1% in our study. This is considerably below the UK population prevalence.