(294) Association Between Initiation of Medical Therapy and Hypo- or Hyperthyroidism - A Population-Based Symmetry Analysis Screening Study

Friday, August 26, 2022

8:00 AM – 6:00 PM CEST

Location: Hall C

Publication Number: 625

Presenting Author(s)

Daniel Pilsgaard Henriksen
– MD PhD., University of Southern Denmark, Department of Clinical Research, University of Southern Denmark, Odense C, Syddanmark, Denmark

Co-Author(s)

Lars Folkestad
– MD, Associate Professor, Department of endocrinology, Odense University Hospital, Denmark
Sara Ali
– Msc Pharm, Department of Clinical Pharmacology, Odense University Hospital, Denmark
Per Damkier, MD, PhD
– Professor, consultant, Department of Clinical Pharmacology, Odense University Hospital, Denmark and Department of Clinical Research, University of Southern Denmark, Denmark, Denmark
Jesper Hallas, MD, PhD, FISPE
– University of Southern Denmark, Odense C, Syddanmark, Denmark

Background: Hypothesis-free screening of symmetry analysis data is a useful approach in adverse drug effect research, helping to detect unknown potential adverse drug effects.

Objectives: The aim of this study was to screen all drugs by using sequence symmetry analysis (SSA) design on population-based large-scale drug and outcome data to detect potential drug-induced hypo- and/or hyperthyroidism.

Methods: The study population was bipartite, and overlapping. It consisted of all adult subjects (>18 years) living on the island of Funen (pop approx. 435 000 in 2016) who 1) Had a diagnosis of hyperthyroidism or hypothyroidism (referred to as the “ICD10-based criteria”) and/or 2) had at least one creatinine measurement as well as a an abnormal TSH measurement (referred to as the “lab-based criteria”).
Subjects were included at the first occurrence of an ICD10-diagnosis of hyper- or hypothyroidism and/or a TSH>4 IU/L and/or TSH < 0.3 (indicating hyperthyroidism).
We then performed a symmetry analysis screening of all newly initiated drug classes (ATC fourth level) one year prior to and after the date of the each of the four diagnostic criteria, anchoring the date to the date of the initial prescription of the given drug class.
We took only SRs into account if the difference between groups, that is persons who filled a prescription before versus after the diagnostic criteria, was >10.

Results: A total of 19,110 individuals with a diagnosis of hypo- or hyperthyroidism, and 73,867 individuals with abnormal TSH measurement were included. Among patients identified by the ICD10-based hyperthyroidism, the strongest signals were macrolides (SR 1.29, 95%CI 1.18-1.42) and betalactamase-sensitive penicillins (SR 1.27, 95%CI 1.16-1.40), whereas sulfur-containing imidazole derivatives (methimazole and carbimazole) (SR 38.47, 95%CI 29.36-52.55), and non-selective beta-blockers (SR 2.62, 95%CI 2.23-3.11) were the strongest signals for ICD10-based hypothyroidism. Among patients identified by lab-based hyperthyroidism, the strongest signals were non-selective (SR 1.78, 95%CI 1.55-2.07), and selective (SR1.46, 95%CI 1.28-1.67) beta-blockers, whereas for sulfur-containing imidazole derivatives (SR 1.96, 95%CI 1.59-2.45) and macrolides (SR 1.43, 95%CI 1.19-1.72) were the strongest signals for lab-based hypothyroidism. As a predefined positive control, we tested amiodarone and lab-based hypothyroidism, and found an SR of 3.39 (95%CI 2.64-4.49).

Conclusions: Using the symmetry analysis to screen a large population-based cohort for laboratory- and diagnosis-confirmed several well-known adverse drug reactions, but also identified unknown but biologically plausible signals. The next step is to explore these more in depth.