Decades of Drug Data Reveal Hidden Trends in Epilepsy Medication Monitoring
A Swedish hospital's analysis of nearly 13,000 carbamazepine tests over 18 years uncovered rising drug concentrations despite declining patient volumes — a pattern labs must account for in quality control. The finding suggests patient-derived benchmarks could improve how hospitals ensure medication safety for epilepsy and psychiatric patients.
Originaltitel: Evaluating Patient Medians for Laboratory Quality Control Using Long-Term Carbamazepine Monitoring Data
Karbamazepinkoncentrationer i patientserum stiger långsamt men stadigt — motsatsen till vad fallande testvolymer skulle förespegla. Uppsala universitetssjukhus analyserade 12 577 blodprover från 3 147 patienter mellan 2007 och 2025, med mediavärden som ökade från 25,2 µmol/L till 31,2 µmol/L över två decennier. Sommarväriken påverkade provtagningsfrekvensen men inte de uppmätta läkemedelsnivåerna. Studien visar att patientmedianer fungerar som ett tillförlitligt komplement till interna kvalitetskontroller — särskilt känslig är metoden för instrumentbyten. Då Uppsala övergick från Architect ci8200 till Cobas Pro c503 2021 visades god överensstämmelse mellan plattformarna (R² = 0,946). För laboratorieledningar och inköpare relevantgörs detta: långtidsdatabaser från terapeutisk läkemedelövervakning möjliggör tidigare upptäckt av analytiska förändringar och stöder evidensbaserat val av analysmetod innan kliniska konsekvenser uppstår.
Abstract Background Carbamazepine is an antiepileptic and mood-stabilizing drug with complex pharmacokinetics and a narrow therapeutic index, making therapeutic drug monitoring (TDM) essential. The therapeutic reference range for carbamazepine at our laboratory is 20–40 µmol/L (4.7–9.4 mg/L). Long-term laboratory data may help identify population trends, seasonal or demographic variation, and the suitability of patient-derived medians as a quality-assurance tool. Methods This retrospective study analyzed 12 577 serum carbamazepine results obtained between May 2007 and December 2025 at Uppsala University Hospital. The total number of unique individuals providing the test results was 3147. Demographic variables (age, sex) and sampling dates were included. Yearly percentiles were used to evaluate long-term trends, and monthly variation was assessed for seasonal effects. Equivalence of analytical results was examined during the transition from the Architect ci8200 to the Cobas Pro c503 platform (February 2021). Results Annual test volume peaked in 2010 (n = 955) and declined to 160 by 2025. Despite reduced testing, carbamazepine concentrations increased steadily over time, with median values rising from 25.2 µmol/L (5.9 mg/L) in 2007 to 31.2 µmol/L (7.4 mg/L) in 2025. Seasonal analysis showed predictable drops in sampling during the Swedish summer vacation period but no meaningful variation in drug concentrations. The method transition in February 2021 showed strong agreement between platforms, with slightly higher values on the Cobas system (Cobas = 1.049 × Architect − 0.164; R2 = 0.946). Conclusions Carbamazepine concentrations in routine patient samples have gradually increased over 2 decades, independent of declining test frequency. Seasonal workload fluctuations did not affect observed levels. Patient-derived medians were stable and sensitive to method changes, supporting their role as complementary internal quality-control tools. Long-term TDM databases provide valuable insights for both clinical interpretation and laboratory quality assurance.