Selecting Representative Medications for Integrated Syndromic Surveillance in Pharmacies in Rural China

M. Yu,Q. Zhao,L. Cheng,L. Palm,W. Yan,X. Song,G. Zhao,B. Xu
DOI: https://doi.org/10.1016/j.ijid.2012.05.337
IF: 12.073
2012-01-01
International Journal of Infectious Diseases
Abstract:Background: Medication sales in pharmacy, as an early indicator of infectious disease outbreaks, were adopted in syndromic surveillance in many countries. From August 1st, 2011 to January 31st, 2012, we monitored 119 medications of 5 categories (39 compound cold medications, 39 cough suppressants, 18 antibiotics, 8 antipyretics and 15 antidiarrheals) for influenza-like illness and gastrointestinal infection from 2 county pharmacies and 7 township pharmacies in rural Jiangxi Province,China. In pharmacies without electronic sales system, sales data was manually recorded and input daily by data collectors (usually the sales assistants) to a web-based platform of an ongoing research project on integrated surveillance system (ISSC). Considering the huge workload of data collection, we plan to reduce the number of monitoring medications through identifying the most frequently purchased medications. Methods: Data were grouped into 3 settings: two for the two county pharmacies, and one for the aggregation of all township pharmacies for its lower sales volume. In each setting, we ranked medications by descending sales volume under every category and selected medications from the top until the sales volume reaches 70% of the category sum. Given the difference in consumer purchasing habits, every finally selected medication should be present at least in two settings. Linear correlation analysis was used to examine the correlation between the selected and the original medications in each category. Results: Totally 26 of the 119 medications including 7 compound cold medications, 9 cough suppressants, 3 antibiotics, 3 antipyretics and 4 antidiarrheals were selected as representative medications, which accounted for 66.2% of the total sales volume of all pharmacies. The Pearson correlation coefficients between the selected and originals for the 5 categories were 0.973(P<0.001), 0.923(P<0.001), 0.908(P<0.001), 0.847(P<0.001), and 0.937(P<0.001) respectively. Conclusion: The selected 26 medications could represent the original 119 medications in accordance with the high correlation. For surveillance in pharmacy in less developed rural area, selecting the most popular and representative medications for reporting is a good solution for reducing workload of data collectors and improving validity of data. The study is financially supported by a grant under the European Union Framework Program 7 (project No: 241900).
What problem does this paper attempt to address?