David J. McLernon, MPhil 1 , Peter T. Donnan, PhD 2 , Stephen Ryder, MD 3 , Paul
Roderick, PhD 4 , Frank M. Sullivan, PhD 2 , William Rosenberg, PhD 5 , John F. Dillon,
FRCP 6
Addresses:
1
Medical Statistics Team, Section of Population Health, Polwarth Building,
University of Aberdeen, Aberdeen, UK.
2 Tayside Centre for General Practice, Community Health Sciences, University of Dundee,
Dundee, UK.
3
Directorate of Medicine, Division of Gastroenterology, Queen's Medical Centre, University
Hospital NHS Trust, Nottingham, UK.
4
Public Health Sciences and Medical Statistics Group, School of Medicine, University of
Southampton, UK.
5
Institute of Hepatology, University College London, UK.
6 Division of Pathology and Neurosciences, Ninewells Hospital and Medical School,
University of Dundee, Dundee, UK.
Correspondence to:
DJ McLernon
Address: Medical Statistics Team,
Section of Population Health,
University of Aberdeen,
Polwarth Building,
Foresterhill,
Aberdeen,
UK,
AB25 2ZD.
Tel: 01224 551878
Fax: 01224 550925
Email: d.mclernon@abdn.ac.uk
Word Count: 2999
1

Abstract
Background: Patients who present with abnormal liver function tests (LFTs) in primary care
2 and no obvious symptoms can be difficult to manage.
Objective: To follow-up a cohort of liver function tested patients to determine their outcome.
Design: Population-based retrospective cohort study.
Setting: Tayside, Scotland from 1989 to 2003.
Subject: Patients with no clinically obvious liver disease at initial liver function testing in primary care.
Main outcomes: Diagnosed liver disease, mortality.
Methods: Record-linkage of databases ascertained risk factors and outcomes. Measures of performance were calculated and Weibull regression analysis from initial LFT date was performed on all outcomes by level of abnormality.
Results: 95,977 patients had 364,194 incident initial LFTs, with median follow-up 3.7 years.
21.7% had at least one abnormal LFT and 1108 (1.15%) developed liver disease. Elevated transaminase was strongly associated with diagnosed liver disease, HR=4.23 (95% CI 3.55,
5.04) for mild levels and HR=12.67 (9.74, 16.47) for severe levels vs. normal. For GGT these hazards were 2.54 (2.17, 2.96) and 13.44 (10.71, 16.87) respectively.
Low albumin was strongly associated with all-cause mortality, HR=2.65 (2.47, 2.85) for mild levels and
HR=4.99 (4.26, 5.84) for severe levels. Sensitivity for predicting events over 5 years was low and specificity high.
Conclusions: All LFTs were predictive markers for liver disease as well as general ill health, although sensitivity was poor. Most patients with abnormal LFTs had no later formal diagnosis of liver disease within the study period. The time taken to develop liver disease in these patients provides opportunity to intervene.
Key words : Liver function tests, survival, mortality, liver disease, sensitivity

Introduction
Liver function tests (LFTs) are widely performed opportunistically in primary and secondary
3 care, even in patients without suspected liver disease, and may trigger further invasive and expensive investigations. When patients present with abnormal liver function tests (LFTs) in primary care with no clinically obvious liver disease, the appropriate level of follow-up to take can be unclear. As a result GPs may over-investigate some and under-investigate others leading to variation in clinical practice. Increasing the number of tests will increase the likelihood of false positive results.
1
However, despite the increasing use of LFTs, patients continue to present with potentially fatal complications of undiagnosed end stage liver disease, which may have been preventable through earlier diagnosis.
Most of the published epidemiological studies report only the prevalence of liver disorders rather than addressing the absolute or relative risks of subsequent liver disease following abnormal LFTs.
2-5
Previous studies have established that ALF may be indicative of neoplasms as well as liver disease and are associated with poor outcome.
6-8
I n many cases this may be due to causes that could be intervened with earlier resulting in reduced morbidity and mortality
.
9
This population-based historical cohort study followed-up all patients, living in Tayside, UK with no clinically recognised liver disease who initially had LFTs undertaken in primary care to subsequent mortality or diagnosed liver disease.

Methods
Study Population
The study population was initially derived from a laboratory database which contained all electronically available LFT results from patients in the Tayside region of Scotland, UK
4 during the fifteen year period from 1989 to 2003. Tayside is a mixed urban/rural region characteristic of Scotland with a population of approximately 410,000. LFTs included bilirubin, albumin, alkaline phosphatase (Alk Phos), gamma-glutamyl transferase (GGT), alanine transaminase (ALT) and aspartate aminotransferase (AST). Since many laboratories only measure either ALT or AST these two tests were combined in this study and are referred to as transaminases.
In brief, patients aged 16 and above with no obvious clinical signs and symptoms of liver disease, with at least two initial LFTs referred from a Tayside GP between 1989 and 2003 were eligible for inclusion. Exclusion criteria detailed elsewhere ensured that the study population of patients had no clinically recognised liver disease at presentation in primary care.
10
Databases
Data were extracted by the Health Informatics Centre (HIC).
11
The databases relevant to this study, listed below, covered the entire study period and were used within procedures approved under the Data Protection Act and Caldicott Guardian.
1.
Death registry from the General Register Office (GRO) for Scotland including causes of death.

5
2.
Carstairs score for deprivation.
12
The score was divided into two categories of affluent and deprived.
3.
Regional biochemistry database. All LFTs came from the largest hospital laboratory in the region, Ninewell’s. Two smaller hospital laboratories contributed electronic
LFT results later, one from 1998 and the other from 2003.
4.
SMR01 database: Holds hospital admissions and procedures for all hospitals in
Scotland. Major comorbidity groups were identified using this database and Diabetes
Audit and Research Tayside Scotland (DARTS) and Heart disease Evidence-based
Audit and Research Tayside Scotland (HEARTS) databases for diabetes and CHD respectively.
13
5.
SMR02 database: Patient records of maternal inpatient and day case episodes allowing identification of women who had their LFTs measured during pregnancy.
6.
SMR04 database: Holds all inpatient and day case episodes for mental health specialties, which identified patients admitted/discharged with diagnoses of alcohol dependency or drug misuse.
7.
Primary care prescription database: Holds encashed prescriptions in Tayside, and allowed identification of patients on some hepatotoxic drugs including statins, antibiotics and NSAIDs.
8.
Epidemiology of Liver Disease In Tayside (ELDIT) database: Contains all Tayside patients with liver diseases that have been ‘electronically’ diagnosed using recordlinkage of biomedical datasets, including virology and immunology.
14,15
All of the electronic databases described above were electronically linked with a unique identifier, the community health index (CHI).
16
The CHI is used for all health encounters in
Tayside for the population registered with a general practice.

6
Outcomes
The primary outcomes following the initial LFT results were all-cause mortality, liver mortality and liver disease diagnosis. For descriptive purposes other outcomes including hospital admission, diagnosis of IHD, cancer, respiratory disease, diabetes and biliary disease were tabulated. The individual liver disease outcomes were also recorded by LFT (Appendix
1).
Statistical Analysis
Normal, mildly elevated and severely elevated categories were defined for each LFT using regional laboratory standard cut-offs which vary by age and sex for some tests (10). Severely elevated was defined as 2.5 times the normal range. Baseline characteristics were tabulated by level of abnormality for each test. These characteristics were age, gender, Carstairs category, comorbidities during the period 1980 to study start (including cancer, diabetes, IHD and respiratory disease and biliary disease), alcohol dependency and drug misuse (from hospitalisations), methadone abuse, pregnancy, and the use of statins, NSAIDs or antibiotics in the three months before LFTs.
Outcomes from the initial LFTs were tabulated. Sensitivity, specificity, positive predictive value (PPV) and negative predictive values (NPV) of the LFTs to diagnose liver disease and to predict death (any cause and liver caused) were calculated over one and five years. Since some patients did not have all their LFTs tested, liver disease diagnosis by LFT testing can be subject to selection bias. The predicted probability of testing for each LFT was found by fitting a logistic regression model adjusted for all the predictors mentioned above.
17,18
The

probability of testing was then used to weight the logistic regression models for predicting
7 outcome.
Survival analysis was conducted to investigate whether abnormality of an initial LFT had an effect on time to the specified outcomes. The starting point was taken as the date of the initial
LFT test and the endpoint was 31 st
December 2003, date of outcome, emigration or death, whichever was earlier. All patients whose endpoint was not the outcome of interest were censored. Weibull regression models were fitted separately for each LFT by level of abnormality adjusted for the baseline characteristics. Initially, a univariate analysis was performed on each of these factors and those with a p-value > 0.3 were excluded from the stepwise regression technique. A multiple imputation technique was used to impute missing values for LFTs.
19 The proportional hazards assumption was checked and survival curves were plotted by initial LFT result. Analyses were performed using SAS (v8) (SAS Institute,
Cary, North Carolina).
Results
Before exclusions, we extracted LFTs from 310,511 patients. When we excluded patients under the age of 16, non-Tayside residents and those who had their initial abnormal LFTs measured in secondary care, 99,165 patients were left. When the remaining exclusion criteria were applied (bilirubin >35 μmol/L i.e. clinically obvious liver disease, complications within
6 weeks and history of liver disease) our study population contained 95,977 patients with
364,194 incident initial LFTs taken in primary care. The median follow-up time was 3.7 years (IQR 1.4, 7.6). 57.9% of patients were female and the median age was 54.6 years (IQR
39.2, 68.8). Alk phos was measured in 99.2% of patients, albumin (99.2%), bilirubin

(93.6%), transaminases (76.5%) and GGT (10.9%). Use of LFTs in primary care increased
8 over time due to a combination of more testing and better laboratory coverage. At initial tests,
21.7% of patients had at least one abnormal LFT.
Baseline characteristics
The mean age of patients with abnormal LFTs ranged from 48.8 to 52.6 years except for albumin which had a mean age of 69.6. Of those with GGT measured, a higher proportion were male and deprived patients – more so than any other LFT (table 1). Abnormal GGT groups also had the highest percentage of patients dependent on alcohol (15.3%), drugs
(0.9%) and methadone (0.7%). Patients with abnormal albumin had much more comorbidity compared to the other LFTs (table 1). The percentage of patients with initially abnormal
LFTs prescribed statins in the three months beforehand was less than those with normal
LFTs. The percentage of patients prescribed NSAIDs or antibiotics in the preceding three months was highest in those with lowered albumin.
Outcomes
Table 2 contains the outcomes per 1000 patient years (TPY) by initial LFT level. Low albumin had much higher rates of mortality from any cause compared to the other LFTs with rates of 166.3 TPY and 260.4 TPY for mildly and severely elevated respectively. Severe alk phos had the next highest death rate of 99.8 TPY. For liver caused mortality, severely elevated GGT had the highest rate of 9.8 TPY followed by alk phos (8.43 TPY) and albumin
(8.19 TPY). Albumin also had the highest rates for cancer diagnosis – for mild and severe they were 56.6 and 96.0 TPY respectively. Transaminase and GGT were most associated with liver diagnoses, with severely abnormal levels having rates of 36.3 TPY and 41.0 TPY

9 respectively. Rates of hospital admission after an abnormal albumin test were extremely high with even a mildly lowered result having a rate over 500 TPY.
Performance measures
For outcomes of all-cause mortality, liver mortality and liver disease diagnosis, GGT had the best sensitivity scores (0.769 and 0.714 for liver mortality over one and five years respectively) (see Table 3). However, all the tests had generally low sensitivity. Albumin had the poorest sensitivity overall however its specificity was the highest (0.980 for liver disease).
All LFTs had NPVs of 0.990 or more for all outcomes except all-cause mortality. All tests had very low PPVs between 0.002 and 0.052 for all outcomes except all-cause mortality, for which albumin had the highest PPV (0.508 within five years).
Survival analysis
Liver disease diagnosis: All the LFTs were significantly predictive of liver disease even after adjusting for risk factors for liver damage (see table 4). Of the mildly elevated LFTs, transaminase had the highest hazard ratio of 4.23 (95% CI 3.55, 5.04) (see Figure 1a). All severely elevated LFTs had hazard ratios over 8, with alk phos highest followed by GGT and transaminase. Other factors predictive of liver disease were older age, deprivation, alcohol dependency, illicit drug use and methadone use. For the transaminase model the hazards for the latter three factors were 4.48 (95% CI 3.70, 5.42), 2.25 (95% CI 1.51, 3.36) and 4.52
(95% CI 3.07, 6.65) respectively.
Liver disease mortality: Of the mildly abnormal LFTs, low albumin had the highest hazard for liver mortality of 7.38 (95% CI 4.60, 11.81). Severely elevated GGT had the highest hazard ratio overall with a value of 25.32 (95% CI 15.27, 41.97), followed by alk phos and

10 albumin (see Table 4). The only baseline factors which predicted liver disease mortality were gender, older age, Carstairs score (deprived) and alcohol dependency. Alcohol dependency had hazards as high as 10.84 (95% CI 7.28, 16.14) for the albumin adjusted model.
All-cause mortality: All LFTs had significantly high hazard ratios for all-cause mortality.
Albumin had the largest hazard ratios (4.99 (95% CI 4.26, 5.84) for severely lowered). For mildly lowered albumin the hazard was over 2.5 times that for a normal albumin (see Figure
1b). Transaminase had the lowest hazards for mortality. The baseline factors in the models predictive of death included gender, age, Carstairs score, all comorbidities, statin, NSAID and antibiotic use, and alcohol, drug and methadone dependency. With the exception of biliary cancer, which had a typical hazard ratio of 15.70 (95% CI 5.06, 48.71), all hazards for these factors were less than two. Statin use was associated with lower risk with a typical hazard ratio of 0.56 (95% CI 0.49, 0.65). All models demonstrated approximate proportional hazards.
Discussion
The most striking observations of this study are that: 1) liver disease is not common in those with abnormal LFTs over a median follow-up time of 3.7 years, 2) GGT shows highest sensitivity for liver disease above other LFTs and is a good predictor of liver disease and liver mortality after adjustment for the bias of selective testing and 3) abnormal LFTs are predictive of death from non-hepatic causes (particularly albumin).
This is the first large-scale population-based analysis of LFTs with a long follow-up period and complete determination of outcome. These results are derived from unselected “real-

world” observations in a geographically defined population. However, liver disease can suffer from under ascertainment as hospital discharge records and death certificates often
11 omit liver disease if it was not the primary cause of death. We have no precise data on the reasons why LFTs were requested and these are likely to include the investigation of suspected liver disease, monitoring of disease or treatment where liver toxicity may occur and for the investigation of non-specific illness. However, we were able to identify patients with a history of major comorbidities since 1980, including cancer, diabetes and IHD using hospital admission records, and patients who were prescribed statins, NSAIDs and antibiotics in the three months before their initial LFTs. The other limitations of electronic data sources are that we have no information on alcohol intake, BMI or other anthropometry associated with
Non-alcoholic fatty liver disease (NAFLD). Scotland has recently developed software which enables extraction of Read coded data from primary care systems which will enable us to address this issue in future prospective studies.
20
However, as regards this study, electronic primary care data would not be available for all patients from 1989 since not all practices would have stored electronic data back then.
Diagnosis of liver disease in the study population was identified throughout the study period using the ELDIT database.
14
ELDIT was created via the record-linkage of various datasets.
These include those SMR hospital admission databases described in the methods, the SMR06 cancer registry which contains all malignant neoplasms and carcinomas in-situ of the liver, and hospital datasets containing virology, immunology, pathology, and endoscopy information.
16 Each liver disease was electronically diagnosed using strict algorithms.
14 15
Since most major liver diseases are diagnosed in secondary care, it is assumed that the
ELDIT database would catch more disease than a primary care clinical database.

We found that the sensitivity of LFTs in detecting liver disease is generally poor whilst in
12 contrast specificity was high. In terms of prediction of future liver disease NPV was high whilst PPV was low. A study in Italy found sensitivity and specificity values for ALT of 0.40 and 0.98 respectively for hepatitis C, which were quite similar to ours for any liver disease.
21
Survival models showed that all the tests have high hazard ratios relating to outcomes of liver disease and mortality from liver disease. Of 667 people who had a severely elevated transaminase over 11% where diagnosed with chronic liver disease, and a mild elevation of the test had a high hazard for developing chronic liver disease of over 4 suggesting that transaminase may be a good predictor.
In Tayside for much of the duration of this study, GGT was not routinely requested as part of the ‘batch of LFTs’. This explains the low numbers of GGTs performed, and it is not surprising that the baseline characteristics of patients for GGT results differed from the other tests. It was the only test requested more frequently for males, illicit drug users and patients living in deprived areas suggesting that GPs selected these patients for GGT testing. The patients not tested for GGT had their GGT results imputed using the ‘gold standard’ multiple imputation technique.
19 Even after this, severely abnormal GGT increases the risk of liver disease over 13 times compared to normal GGT and 2.5 times if mildly abnormal. This suggests strongly that GGT provides additional information over the other LFTs and should be considered as an important and informative part of the LFTs. In light of this finding the practice by some laboratories of not routinely measuring GGT should be reviewed.
Why is the sensitivity of LFTs for liver disease so poor? Of those who subsequently develop liver disease their first LFT may be normal. The study in Korea found that patients with slightly raised transaminases, which were still in the normal range, developed liver disease

and suggests an adjustment of the normal limit.
7
Also many patients with abnormal LFTs
13 detected in this study may have had no subsequent formal diagnosis of liver disease because of a lack of investigation and a limited time interval to develop complications. So that it is possible that there is a pool of undiagnosed liver disease within this cohort. It is likely on epidemiological grounds, that the majority of these abnormalities could be attributable to undiagnosed alcohol related liver disease, Hepatitis C or NAFLD. The fact that this group of patients did not come to harm during the study is reassuring. However, the study follow-up period is medium term (median of 3.7 years) compared to the natural history of these diseases. It does however illustrate a window of opportunity to intervene in these patients with life style advice, alcohol intake reduction and therapies for drug abuse.
All-cause mortality is a much more common outcome in this study than liver disease. This suggests that these tests may be better markers of poor health than liver disease, thus possibly justifying the increasing use of LFTs as a screen for general illness. In particular reduced albumin levels are associated with serious illness,
22-24
as substantiated by the fact that even a mildly reduced albumin level has a hazard of mortality over two and a half times that of the normal. This raises uncertainty over what the most appropriate investigation for patients with abnormal LFTs is if most do not have underlying liver disease but have increased risks of several other diagnoses e.g. cancer and cardiovascular disease. Indeed, it may be the case that undiagnosed underlying liver disease is a risk factor for death from other causes.
In summary, this study has described the epidemiological association of abnormalities in an initial primary care panel of LFTs with important health outcomes. Until now the strength of the association with death or liver disease in patients with abnormal levels of LFTs was not known. Further work is planned to develop more predictive models for clinical decision

14 support and conduct cost-effectiveness analyses on LFTs referred from primary care to ascertain the optimal management strategies that will reduce costs to the health service and optimize patient care.
Acknowledgements : The authors wish to thank Dr Ian Kennedy from the biochemistry department at the University of Dundee for providing some of the biochemistry data and invaluable information on it.
Ethical approval: Local research ethics committee approval was secured for the study.
Funding : This work was supported by the National Health Service Research & Development
Programme Health Technology Assessment Programme (project number 03/38/02).
Department of Health Disclaimer: The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the Department of Health.
Conflict of interests : None.
References
1. Van Walraven C, Naylor CD. Do we know what inappropriate laboratory utilization is? A systemic review of laboratory clinical audits. JAMA. 1998 1998; 280: 550-558.
2. Mann RE, Smart RG, Govoni R. The epidemiology of alcoholic liver disease. Alcohol
Res Health. 2003 2003; 27(3): 209-219.
3. Watson JP, Brind AM, Chapman CE, et al. Hepatitis C virus: epidemiology and genotypes in the north east of England. Gut. 1996 1996; 38: 269-276.

15
4. Bellentani S, Tiribelli C, Saccoccio G, et al. Prevalence of chronic liver disease in the general population of northern Italy: the Dionysos study. Hepatology. 1994 1994; 20: 1442-
1449.
5. Fatatsuka M, Kitano T, Nagano M, et al. An epidemiological study with risk analysis of liver disease in the general population living in a methyl mercury polluted area. J
Epidemiol Community Health. 1992 1992; 46: 237-240.
6. Sorensen HT, Moller-Petersen JF, Felding P, Andreasen C, Nielsen JO. Epidemiology of abnormal liver function tests in general practice in a defined population in Denmark. Dan
Med Bull. 1991 1991; 38: 420-422.
7. Kim HC, Nam CM, Jee SH, Han KH, Oh DK, Suh I. Normal serum aminotransferase concentration and risk of mortality from liver diseases: Prospective cohort study. BMJ. 2004
4/24/2004;328(doi/10.1136/bmj.38050.593634.63).
8. Roderick P. Commentary: Liver function tests: defining what's normal. BMJ. 2004
2004;328:987.
9. Sherwood P, Lyburn I, Brown S, Ryder S. How are abnormal results for liver function tests dealt with in primary care? Audit of yield and impact. BMJ. 2001 2/3/2001; 322: 276-
278.
10. Donnan PT, McLernon D, Steinke D, et al. Development of a decision support tool to facilitate primary care management of patients with abnormal liver function tests without clinically apparent liver disease.[HTA03/38/02] Abnormal Liver Function Investigations
Evaluation (ALFIE). BMC Health Services Research. 2007; 7(54).
11. The Health Informatics Centre. http://www.dundee.ac.uk/hic . 2007.
12. Carstairs V, Morris R. Deprivation and mortality: an alternative to social class?
Community Med. 1989; 11(3): 210-219.

16
13. Morris AD, Boyle DI, MacAlpine R, et al. The diabetes audit and research in Tayside
Scotland (darts) study: electronic record linkage to create a diabetes register. BMJ. 1997
August 30, 1997; 315(7107): 524-528.
14. Steinke DT, Weston TL, Morris AD, MacDonald TM, Dillon JF. The epidemiology of liver disease in Tayside database: a population-based record-linkage study. J Biomed
Inform. 2003; 35: 186-193.
15. Steinke DT, Weston TL, Morris AD, MacDonald TM, Dillon JF. Epidemiology and economic burden of viral hepatitis: an observational population based study. Gut. 2002; 50:
100-105.
16. Evans JMM, MacDonald TM. Record-linkage for pharmacovigilance in Scotland. Br
J Clin Pharm. 1999; 47:105-110.
17. Begg CB, Greenes RA. Assessment of diagnostic tests when disease verification is subject to selection bias. Biometrics. 1983; 39: 207-215.
18. McLernon DJ, Donnan PT, Sullivan F, Gray M, Weller D. Receiver operating characteristics of the Prostate Specific Antigen test in an unselected population. J Med
Screen. 2006; 13(2): 102-107.
19. Rubin DB. Multiple imputation for nonresponse in surveys. New York: Wiley; 1987.
20. Scottish Enhanced Functionality. http://www.sef.scot.nhs.uk/ . Accessed Feb 2009
21. Prati D, Taioli E, Zanella A, et al. Updated definitions of healthy ranges for serum alanine aminotransferase levels. Ann Intern Med. 2002; 137: 1-9.
22. Schalk BWM, Visser M, Bremmer MA, Penninx BWJH, Bouter LM, Deeg DJH.
Change of serum albumin and risk of cardiovascular disease and all-cause mortality:
Longitudinal Aging Study Amsterdam. Am J Epidemiol. 2006; 164(10): 969-977.

23. Soriano S, Gonzalez L, Martin-Malo A, Rodriguez M, Aljama P, Cabrera SS. Creactive protein and low albumin are predictors of morbidity and cardiovascular events in
17 chronic kidney disease (CKD) 3-5 patients. Clin Nephrol. 2007; 67(6): 352-357.
24. Omran ML, Morley JE. Assessment of protein energy malnutrition in older persons, part II: laboratory evaluation. Nutrition. 2000; 16: 131-140.

18
Table 1 Baseline and historical characteristics of the population (n=95977) by LFT (normal/abnormal)
Baseline
Characteristics
Age (mean (SD))
Male
Carstairs category
Affluent
Deprived
Comorbidity
Ischaemic HD
Diabetes
Respiratory
Cancer 1
Biliary disease
Medication in previous 3 months
Statins
NSAIDs
Antibiotics
Abusive substance
Alcohol
Drug
Methadone
Population
(%)
(N=95977)
Albumin (%)
Normal
(n=93240)
Abnormal
(n=1947)
Alkaline phosphatase
Normal
(n=85329)
(%)
Abnormal
(n=9928)
Transaminase (%)
Normal
(n=68314)
Abnormal
(n=5107)
GGT (%)
Normal
(n=8861)
Abnormal
(n=1623)
Bilirubin (%)
Normal
(n=83457)
Mildly elevated
(n=6365)
53.8 (18.6) 53.5(18.6) 69.6 (18.5) 54.5(19.1) 48.8 (14.8) 53.1(18.9) 52.6(18.2) 50.7(17.4) 52.5(15.5) 54.0(18.6) 52.6(20.0)
42.1
49.4
50.6
41.8
49.4
50.6
38.3
49.5
50.5
43.1
42.6
57.4
30.1
50.2
49.8
42.6
51.4
48.6
59.1
50.2
49.8
55.6
38.6
61.4
65.7
39.7
60.3
41.8
49.0
51.0
54.1
53.7
46.3
5.6
1.4
2.8
3.8
1.8
3.3
7.0
8.7
2.7
0.4
0.4
5.6
1.4
2.7
3.7
1.8
3.4
7.0
8.6
2.7
0.4
0.4
2.5
2.4
6.0
5.4
9.3
0.8
9.3
14.3
2.8
0.2
0.6
6.0
1.5
2.8
4.0
1.7
3.6
6.9
8.6
2.5
0.3
0.4
1.2
2.1
2.7
2.2
2.4
1.2
8.4
9.1
4.4
0.7
0.5
5.9
1.6
2.8
4.0
1.7
4.3
4.6
8.5
2.5
0.4
0.4
1.7
1.8
4.4
2.9
3.5
3.2
5.4
8.9
5.4
0.5
0.5
1
Not including biliary cancer (comes under biliary disease) or hepatocellular cancer (comes under liver disease)
1.9
5.1
8.3
5.6
0.7
0.5
4.2
0.9
2.5
3.6
1.6
4.4
0.9
2.0
2.7
2.0
2.0
6.7
9.1
15.3
0.9
0.7
3.5
7.2
8.9
2.6
0.4
0.4
5.8
1.4
2.7
3.8
1.8
2.6
4.4
7.1
3.9
0.2
0.2
4.7
1.4
2.2
2.6
1.4

Table 2 Outcome rates by first LFT level of severity per 1000 person years (TPY)
Outcomes
Death
Any cause
Liver
Cancer a
Biliary cancer
Liver diagnosis
Hospital admission
Other diagnosis
Ischaemic HD
Renal disease
Diabetes
Respiratory
Stroke
Cancer a
Biliary disease
Biliary cancer
Population
(N=95977)
26.60
0.50
3.69
0.40
2.37
Normal
(n=93240)
24.78
0.44
3.45
0.38
2.24
Albumin
Mildly lowered
(n=1710)
166.26
4.47
21.62
1.30
9.70
Severely lowered
(n=237)
260.41
8.19
32.76
1.64
Normal
(n=85329)
26.88
0.37
3.66
0.38
26.01 1.88
ALP
Mildly elevated
(n=9598)
23.83
1.27
3.76
0.44
5.35
Transaminase
Severely elevated
(n=330)
99.75
8.43
16.16
2.81
Normal
(n=68314)
22.96
0.31
4.94
0.34
31.45 1.84
Mildly elevated
(n=4440)
31.66
2.43
5.46
0.67
10.42
Severely elevated
(n=667)
41.19
4.08
7.34
2.04
36.33
165.93 162.91 531.54 1149.98 166.95 159.20 393.26 170.71 187.22 264.21
10.68
3.58
4.61
10.75
6.20
12.63
3.39
0.50
10.59
3.39
4.57
10.31
6.04
12.10
3.39
0.48
21.92
19.18
11.23
49.70
21.41
56.68
5.70
1.31
21.77
27.12
3.28
58.98
22.01
96.01
8.45
1.64
10.96
3.61
4.52
10.85
6.39
12.55
3.12
0.48
9.07
3.53
5.45
10.22
4.95
12.77
5.09
0.47
7.94
5.02
8.79
21.30
8.56
55.89
26.47
3.52
9.39
3.03
4.17
9.48
5.33
12.16
2.98
0.45
10.59
3.79
9.28
11.23
7.17
13.80
5.90
0.61
12.62
5.73
5.80
14.30
5.76
13.28
22.34
3.27
19

Outcomes
Normal
(n=8861)
GGT
Mildly elevated
(n=1094)
Severely elevated
(n=529)
Bilirubin
Normal
(n=83457)
Death
Any cause
Liver
Cancer a
Biliary cancer
Liver diagnosis
Hospital admission
20.52
0.39
4.09
0.33
2.26
149.35
33.38
2.41
6.12
0.56
10.51
183.59
54.40
9.82
5.73
2.86
40.95
262.36
26.38
0.42
3.68
0.39
2.22
165.23
Other diagnosis
Ischaemic HD
Renal disease
Diabetes
Respiratory
Stroke
Cancer a
Biliary disease
Biliary cancer
8.65
2.44
3.09
9.04
4.80
10.36
2.83
0.42
9.55
3.73
5.26
10.82
4.31
17.62
3.56
0.74
9.23
5.37
8.36
18.16
7.47
18.33
15.50
3.28
1
Not including biliary cancer or hepatocellular cancer (comes under liver disease)
10.87
3.60
4.52
10.78
6.19
12.68
3.33
0.48
Mildly elevated
(n=6365)
31.06
1.77
3.58
0.46
4.72
166.83
9.34
3.21
5.79
10.86
7.20
12.19
4.73
0.71
20

Table 3 Percentage sensitivity, specificity, PPV and NPV of LFTs at the first GP consultation in patients with no obvious liver disease for 1 and 5 year periods
LFT
Alk Phos
Performance measure
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
All death
1 yr
17.4
89.8
4.9
97.3
5 yrs
11.2
89.7
9.7
91.0
Liver death
1 yr
48.8
89.6
0.2
99.9
5 yrs
37.6
89.6
0.5
99.9
Albumin Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
Transaminase Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
GGT Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
20.4
99.8
29.4
98.0
11.5
93.2
4.5
97.3
37.1
85.3
7.4
97.5
11.4
98.9
50.8
91.8
9.5
93.3
11.0
92.0
27.4
85.7
15.2
91.2
26.8
98.0
0.6
99.9
37.5
93.1
0.2
99.9
76.9
84.6
0.4
99.9
19.2
98.0
1.2
99.9
40.2
93.1
0.7
99.9
71.4
84.7
1.0
99.9
Bilirubin 1 Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
10.5
93.0
4.3
97.2
8.5
93.1
11.0
91.0
35.9
92.9
0.2
99.9
24.8
92.9
0.5
99.9
1 The sensitivity and ppv for bilirubin only includes patients with bilirubin ≤35 at initial GP consultation.
Note: All performance measures were adjusted for the probability of testing.
17
Liver disease
1 yr
42.5
89.7
1.5
99.8
5 yrs
32.1
89.8
2.5
99.4
15.0
98.0
2.6
99.7
42.1
93.2
2.2
99.8
72.4
85.1
3.1
99.8
16.8
92.9
0.9
99.7
14.9
93.0
1.7
99.2
6.6
98.0
3.5
99.0
35.8
93.3
3.9
99.5
61.9
85.4
5.2
99.4
21

22
Table 4 Weibull regression model results by level of initial LFT abnormality (vs normal) for time to liver disease, all cause mortality and liver disease mortality
Variable
Albumin
Mildly lowered
Severely lowered
Alkaline phosphatase
Mildly elevated
Severely elevated
Transaminase 1
Mildly elevated
Severely elevated
GGT 1
Mildly elevated
Severely elevated
Bilirubin 2
Mildly elevated
Liver disease All cause mortality
Hazard ratio (95% CI) Hazard ratio (95% CI)
3.41 (2.55, 4.55) 2.65 (2.47, 2.85)
8.48 (5.04, 14.28) 4.99 (4.26, 5.84)
2.91 (2.49, 3.39)
14.42 (10.20, 20.37)
1.80 (1.69, 1.91)
2.88 (2.44, 3.40)
4.23 (3.55, 5.04)
12.67 (9.74, 16.47)
1.35 (1.26, 1.44)
1.88 (1.58, 2.23)
1.56 (1.48, 1.63) 2.54 (2.17, 2.96)
13.44 (10.71, 16.87)
2.02 (1.68, 2.44)
2.90 (2.61, 3.23)
1.20 (1.12, 1.29)
Liver disease mortality
Hazard ratio (95% CI)
7.38 (4.60, 11.81)
16.17 (6.41, 40.81)
3.81 (2.72, 5.32)
17.81 (9.20, 34.49)
5.41 (3.80, 7.71)
7.17 (3.75, 13.70)
4.89 (3.43, 6.99)
25.32 (15.27, 41.97)
3.89 (2.76, 5.48)
Notes: Weibull models were fitted separately for each LFT and were adjusted for all baseline factors described in the text.
1
The missing data from transaminase and GGT were imputed using a multiple imputation method for the modelling.
2
Patients with severely elevated bilirubin were omitted from the population and thus this cohort is not in the survival model for bilirubin.

Captions to Figures
Figure 1. Survival by level of abnormality from: a) first transaminase test to liver disease
23 diagnosis; b) first albumin test to all cause mortality