HiCorr: an influenza surveillance model using over-the-counter sales from community pharmacies

Abstract Background In large-scale community transmission, such as seasonal influenza, monitoring geographic trends and estimating the transmission intensity are critical to support public health decision-making. Previous studies have successfully used sales data from over-the-counter (OTC) products in community pharmacies to detect and monitor different epidemiological outbreaks, developing surveillance systems that are able to anticipate the load on primary care services and hospitals by up to 3 weeks. Methods The subset of OTC products sold in Portuguese community pharmacies was selected by correlating sales data to primary care attendance in previous flu seasons. This data pertained to daily coded episodes of influenza-like-illness (International Classification of Primary Care, 2nd Edition, code R80) in the Portuguese health cluster Oeste Sul. By fitting a moving epidemic method to historical sales data and then applying it to daily collected sales data, the developed index was used to anticipate the relative load to the primary care system as well as the start and the peak of epidemic activity for the current flu season. Results were compared to the public health data available on respiratory infections in Portugal. Results The model pinpointed the onset of the 2023-2024 flu epidemic in week 47 and its peak in week 52. The index curve showed a high correlation with the data published by the Portuguese national health service, with no time lag. However, the availability of consolidated public data was delayed, on average, by approximately two weeks. Discussion and conclusions Data from community pharmacies significantly enhance the early detection of seasonal flu trends, while reflecting population distribution. This underscores their invaluable contribution to public health assessments and the potential for improving the timeliness and accuracy of seasonal flu surveillance. Key messages • Community pharmacy data significantly enhance the early detection of seasonal flu trends. • Real world sales data maybe useful to support epidemic surveillance in public health.