An ecological time series study of lagged associations between pesticide use and cancer incidence in Ukraine
TL;DR
Pesticide use in Ukraine demonstrated wave-like dynamics, with a peak in 2012, a subsequent decline in 2014–2020, and a moderate increase in 2021, whereas herbicides predominated throughout the study period, suggesting sensitivity to shared temporal trends.
Abstract
Introduction Cancer remains a leading cause of mortality and health loss worldwide, and the use of pesticides as potential environmental determinants of oncological risk is being increasingly discussed. Materials and methods This ecological time series study assessed the associations between pesticide use in Ukraine (2000–2021) and cancer incidence (2014–2021) using national aggregated data. Cancer-incidence data were obtained from the National Cancer Registry of Ukraine, and pesticide use data were obtained from FAOSTAT and the State Statistics Service of Ukraine. Analyses included total pesticide use and major pesticide groups (fungicides/bactericides, herbicides, insecticides, plant growth regulators, and other pesticides). Given the potential impact of the COVID-19 pandemic on cancer detection and registration during 2020–2021, two analytical scenarios were applied (2014–2021 and 2014–2019). Associations were assessed using Spearman lag correlations (0–14 years) with Benjamini–Hochberg false discovery rate correction. Additional sensitivity analyses included first differences and linear detrending. All analyses were considered hypothesis-generating because of the short time series and ecological study design. Results Pesticide use in Ukraine demonstrated wave-like dynamics, with a peak in 2012, a subsequent decline in 2014–2020, and a moderate increase in 2021, whereas herbicides predominated throughout the study period. For total cancer incidence, lag correlations varied substantially by time window and were largely attenuated after detrending and differencing, suggesting sensitivity to shared temporal trends. Short-lag inverse associations were interpreted cautiously because they likely reflected temporal structure rather than immediate biological effects. In site-stratified analyses, partly reproducible positive ecological association patterns within the 5–11-year lag window were observed for female breast cancer, female and male colon cancer, and female trachea/bronchus/lung cancer. The most consistent associations involved herbicides, fungicides/bactericides, insecticides, total pesticide use, and area-based pesticide-use indicators. Conclusions These findings are hypothesis-generating and do not support causal inference. Because of the ecological design, ecological fallacy and residual confounding cannot be excluded. Longer time series, refined exposure characterization, and analytical approaches capable of accounting for temporal structure and delayed effects are needed.