How are abnormal results for liver function tests dealt with in primary care? Audit of yield and impact

Introduction

Standard laboratory tests for liver function are often requested by doctors in patients with non-specific symptoms such as tiredness, abdominal pain, dyspepsia, and weight loss. Considerable uncertainty exists as to the appropriate follow up of mildly abnormal results in patients with no signs suggestive of liver disease, and there is a widespread assumption by both patients and doctors that all abnormal liver biochemistry is due to alcohol excess. Few data are currently available as to the yield of investigating mildly abnormal liver biochemistry in such a group. We aimed to follow up all abnormal results for liver function that were greater than twice the upper limit of normal from our clinical chemistry laboratory for a period of six months, requested by doctors serving around 50% of the population of Nottingham, to determine the underlying diagnosis.

Subjects and methods

We identified all adult patients from the Queen's Medical Centre who had an abnormal result for liver function based on a test requested by their doctor between 1 January and 30 June 1995, from the computer database of the clinical chemistry department. The results of liver function tests were defined as abnormal if any one of the concentrations of γ-glutamyl- transferase, alanine aminotransferase, or alkaline phosphatase were at least twice the upper limit of the reference range. A median of 15 months later (range 12-21) we examined the hospital's computer databases for biochemical, virological, immunological, and radiological investigations, outpatient hospital attendances, and inpatient stays for these patients. We excluded from further analysis those patients who were under recent hospital follow up by a physician, oncologist, or general surgeon on the assumption that the abnormal results for liver function were part of the patients' current clinical problems.

Patients who did not fall into these categories were further investigated. Firstly, one of the investigators (PS or SR) visited each patient's doctor and examined the patient's notes, at least 12 months after the index abnormal result. Patients who had had no or incomplete investigations were then invited by letter from their doctor to attend for further investigation. An appointment for a liver clinic was requested if appropriate.

Ethics committee approval for this study was not sought. This was because further investigation of the index abnormal result for liver function, including examination of all relevant available medical records, was thought to be clinically indicated. Patients were offered further investigation on a voluntary basis after full consultation.

Results

Of 8208 tests for liver function requested by doctors in the six month period, we identified 933 patients with abnormal results that were twice the upper limit of the local reference range. Full follow up data were obtained in 873 (93.6%). Of the 933 patients, 497 (53.3%) had at least one result for liver function that was three times the upper limit of the local reference range, and 142 (15.2%) had at least one result that was four times the upper limit of the local reference range.

Tests had been requested for follow up of known liver disease in 531 patients, and the patients were already under hospital review. We examined the doctors' notes for the remaining 342 patients who were not under follow up at hospital (table). Of the 157 (46%) of these patients who were thought to require further investigation, no further tests had been requested for 91 (58%) and the other 66 had been partially investigated by their doctor: tests included autoantibody screen (28 patients), serology for hepatitis B (29) and C (22), serum ferritin concentration (10), and ultrasonography of the abdomen (18).

For seven patients a result suggesting a treatable chronic liver disease had been returned but not followed up, and no referral had been made to a gastroenterologist (table). All 157 patients were invited for repeat liver function tests, and, if required, further investigation was undertaken by the hospital's liver clinic. Overall, 101 (64%) patients had a liver biopsy, providing a diagnosis in 81 when serological tests gave normal results. Of these 157 patients, 97 (62%) had an identifiable diagnosis requiring hospital intervention or follow up, or both; eight of these had a viral hepatitis that was communicable and five had an inherited chronic liver disease (table). Of the 195 patients who had liver function tests repeated on request of their doctor or hospital, 74 (38%) showed spontaneous resolution of the abnormal test result.

Discussion

Commonly, liver function tests are transiently abnormal after acute alcohol excess, minor viral illness, or a drug reaction.1 Liver enzyme concentrations that are chronically increased can result from chronic high alcohol consumption, obesity (particularly in men), and smoking (in women). 1 2 The pattern of changes in liver function tests can suggest a diagnosis, but even large increases are non-specific,3 and liver biopsy is often required for a firm diagnosis.4^–6 As standard liver function tests are an inexpensive investigation, the first step towards discovery of an abnormal test result should be a repeat test after several weeks of alcohol abstention if no other clinical diagnostic clues are apparent. In our study the abnormal test result resolved spontaneously in 38% of patients.

We defined “abnormal” test results as those that were twice the upper limit of the local reference range. This arbitrary cut off point was chosen simply because an initial survey showed six times the number of abnormal test results if a more minor abnormality was considered; well beyond our means of investigation. Our study was not designed to define the sensitivity and specificity of liver function tests in primary care. Major chronic liver disease can exist with normal or near normal test results, and using our sampling method will clearly miss disease. In a study from the era before hepatitis C, Van Ness examined patients referred with one or more results for liver enzyme concentrations that were 1.5 times the upper limit of normal. He found that only 6 of 90 patients had normal sample results obtained by liver biopsy.5

Persistently abnormal test results for liver function require investigation as they can be associated with life threatening potentially communicable yet treatable diseases, as our study shows. Some of the more important diagnoses can be indicated by a relatively inexpensive battery of serological tests, comprising hepatitis B surface antigen, hepatitis C antibody, an autoantibody screen, and concentrations of immunoglobulins and ferritin. Early diagnosis of chronic viral hepatitis, haemochromatosis, autoimmune hepatitis, and possibly primary biliary cirrhosis can improve prognosis.

Many of the patients not investigated had been considered by their doctor, often correctly, as having alcoholic liver disease. It is important to note that alcoholics can have other liver diseases—for example, hepatitis B and C,7^–9 and previous studies have shown that liver biopsy can detect unsuspected liver diseases in patients with a known high alcohol consumption.4 Furthermore, abnormal results for liver function in people who drink excess alcohol, even in those with normal synthetic function, may imply major liver disease, including cirrhosis. Many patients with alcohol problems may be referred to a multidisciplinary alcohol team. A physical assessment of patients with alcohol problems can help. Knowledge of the severity of a physical illness—for example, with biopsy staging of alcoholic liver disease—may impact on behaviour when other advice has failed. Medical complications of alcohol excess can also be anticipated and partly ameliorated.

The nature of our study's design meant that it highlighted instances where a diagnosis had been delayed, in particular in the seven patients who had a “diagnostic” test result but were not referred for treatment. Of 873 patients, 157 (18%) were not appropriately followed up and 97 (11%) with abnormal test results over six months had undetected major liver disease and would probably have benefited from follow up at hospital.

One strategy to improve follow up of abnormal test results would be to issue test reports that suggest further investigations, but as results should be interpreted within the clinical context this may be misleading, and to our knowledge no prospective evaluation of such a policy has been reported. Most studies using guidelines to attempt to change behaviour regarding requesting blood tests have been aimed at reducing usage of pathology services rather than avoiding missing treatable diagnoses, often with considerable success. 10 11 In a US study of supplementary blood tests in diabetic patients in primary care, less than 20% of patients received the recommended number of tests.12 We aim to complete the audit cycle by reaudit after the publication of appropriate guidelines.