COVID-19 and pharmaceutical sales in Karaman, Turkey (2019–2021): a hierarchical ATC analysis

Background The COVID-19 pandemic altered pharmaceutical consumption, but most analyses aggregated sales at broad therapeutic levels, masking category-level variation. Evidence from middle-income countries remains limited.Research design and methods This retrospective study analyzed pharmaceutical sales from 10 community pharmacies and one hospital pharmacy unit in Karaman, Turkey (2019–2021). Products were consolidated using ATC code, yielding 552 products across five ATC levels (m = 14–552). Paired t-tests, blocked and unrestricted ANOVA, Friedman tests, and Wilcoxon comparisons were performed with Benjamini–Hochberg (BH) and Bonferroni corrections.Results Aggregate sales declined significantly at Levels 2–5 in 2020 (BH-adjusted-p = 0.014–0.042), concentrated in the first half of 2020, not persisting into 2021. Effect sizes declined with depth and the period effect was undetected in unrestricted ANOVA but significant with blocking (p = 0.017). Nine categories declined; only antidiabetics increased (A10BD; BH-adjusted p = 0.026). Kendall’s W declined with depth (0.735 → 0.077).Conclusions The pandemic was associated with a broad decline in pharmaceutical sales. Hierarchical ATC analysis revealed that aggregate patterns masked opposing movements. Monitoring sales at multiple ATC levels during emergencies is recommended. Limitations include non-random sampling, exclusion of online and hospital medications, and the inability to establish causality.