Abstract and Introduction
Background & Aims Drug induced liver injury (DILI) is largely underreported, leading to underestimation of its burden. Electronic detection of DILI in healthcare databases shows promise to overcome the issues of spontaneous reporting. The performance of detection algorithms may vary because of inconsistent DILI definition and detection criteria. We performed a systematic review and meta–analysis to identify the DILI detection criteria used in health record databases and determine the performance characteristics of the detection algorithms.
Methods We searched PubMed, EMBASE and Scopus for studies that utilized laboratory threshold criteria to identify DILI cases. Validation studies were included in the meta–analysis. Data were abstracted using standardized forms and quality was assessed using modified Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS–2) criteria. We evaluate the performance characteristics of the detection algorithm by obtaining the pooled estimate of the positive predictive value (PPV) assuming a random effects model.
Results A total of 29 studies met the inclusion criteria for the systematic review; 25 of these studies (n = 35 948) had PPV estimates for performing the meta–analysis. The PPV of DILI detection algorithms was low, ranging from 1.0% to 40.2%, with a pooled estimate of 14.6% (95% CI 10.7–18.9). Algorithms that performed better had prespecified exclusion diagnoses as well as drugs of interest to minimize false positives.
Conclusion Algorithm performance varied with different case definitions of DILI attributed to different laboratory threshold criteria, diagnosis codes, and study drugs.
Drug induced liver injury (DILI) is an unexpected adverse hepatic reaction on the basis of the pharmacological action of the given medication at recommended dosage. DILI commonly refers to idiosyncratic DILI and is distinct from liver injury secondary to drug overdose.[1,2] Definitions of suspected agents in DILI vary from synthetic drugs, complementary and alternative medicine, herbs and dietary supplements; with the latter products being better classified as herb–induced liver injury. Estimations of population–based incidence rates ranged from 13.9 cases per 100 000 inhabitants in France to 19.1 cases per 100 000 inhabitants in Iceland. Rates of DILI are higher in hospitalized patients, ranging from 0.12 to 1.4 per 100 admissions.[5,6] Although DILI is rare, the consequences could be severe. In the US, 13% of acute liver failure (ALF) cases are caused by idiosyncratic DILI, with more than half of the cases requiring liver transplantation. Moreover, DILI is the top reason for market withdrawal of medication.[8,9]
Traditionally, signals of DILI are detected via spontaneous reporting. However, this passive surveillance method is limited by under–reporting. As DILI is commonly diagnosed according to exclusion criteria and clinical likelihood based on prior reports of the incriminating drug causing DILI in other patients, the low reporting inadvertently results in many cases of DILI remaining unrecognized. Sgro et al suggested that less than 6% of hepatic adverse reactions are reported. Prospective studies could provide good quality data to confirm diagnosis of DILI but these studies are resource intensive. The advent of electronic medical records (EMR) in healthcare systems provides opportunity to harness the power of data mining via rule–based or machine learning algorithms to detect and build catalogues for DILI. Several studies have looked into detection of adverse drug events (ADE) using routinely collected data such as diagnosis codes and laboratory parameters.[11–13] Nonetheless, this is particularly challenging for DILI because competing aetiologies mimicking DILI have to be ruled out via a diagnostic algorithm that incorporates careful physical examination, history taking and a series of serological and imaging tests.[1,14,15] Hence, detection of DILI requires substantial effort because it remains a diagnosis of exclusion.[14,16]
A definitive clinical diagnosis of DILI can be difficult to establish, as a gold standard diagnostic test, such as a highly specific biomarker, is currently unavailable.[15,17] Although serum aminotransferases remain the hallmark for detecting liver injury, signals based on abnormal liver function tests may be confounded by underlying diseases.[15,18] Furthermore, there is a lack of consensus on which laboratory threshold criteria should be used to identify DILI cases. For example, three different national DILI registries in Spain, Sweden and the USA use different cut–off values for their inclusion criteria.[20–22] However, as suggested by Aithal et al, DILI cannot be merely defined by the rise in serum transferases and requires causality assessment that incorporates temporal relationship with medication exposure, pattern of liver injury and exclusion of alternative diagnoses to confirm cases. Studies of different causality assessment methods have compared between the Roussel Uclaf Causality Assessment Method (RUCAM), Maria and Victorino scale, Digestive Disease Week–Japan (DDW–J) scale, and structured expert opinion, but there is no internationally accepted standardized causality assessment method used to diagnose DILI.
The variation in definition of DILI criteria and causality assessment used could result in a drastic difference in the performance of DILI detection algorithms in health record databases. Therefore, for efficiency of future research involving DILI detection algorithms, it would be imperative to elicit the better performing algorithms and delve into characteristics associated with their performance. The primary objectives of this systematic review are (i) to identify the DILI detection criteria used in health record databases and (ii) to determine the performance characteristics of the detection algorithms via meta–analysis.
Liver International. 2018;38(4):742-753. © 2018 Blackwell Publishing