Tracking Cholera through Surveillance of Oral Rehydration Solution Sales at Pharmacies: Insights from Urban Bangladesh

Abstract

BackgroundIn Bangladesh, pharmacy-purchased oral rehydration solution (ORS) is often used to treat diarrhea, including cholera. Over-the-counter sales have been used for epidemiologic surveillance in the past, but rarely, if ever, in low-income countries. With few early indicators for cholera outbreaks in endemic areas, diarrhea-related product sales may serve as a useful surveillance tool.Methodology/Principal FindingsWe tracked daily ORS sales at 50 pharmacies and drug-sellers in an urban Bangladesh community of 129,000 for 6-months while simultaneously conducting surveillance for diarrhea hospitalizations among residents. We developed a mobile phone based system to track the sales of ORS and deployed it in parallel with a paper-based system. Our objectives were to determine if the mobile phone system was practical and acceptable to pharmacists and drug sellers, whether data were reported accurately compared to a paper-based system, and whether ORS sales were associated with future incidence of cholera hospitalizations within the community. We recorded 47,215 customers purchasing ORS, and 315 hospitalized diarrhea cases, 22% of which had culture-confirmed cholera. ORS sales and diarrhea incidence were independently associated with the mean daily temperature; therefore both unadjusted and adjusted models were explored. Through unadjusted cross-correlation statistics and generalized linear models, we found increases in ORS sales were significantly associated with increases in hospitalized diarrhea cases up to 9-days later and hospitalized cholera cases up to one day later. After adjusting for mean daily temperature, ORS was significantly associated with hospitalized diarrhea two days later and hospitalized cholera one day later.Conclusions/SignificancePharmacy sales data may serve as a feasible and useful surveillance tool. Given the relatively short lagged correlation between ORS sales and diarrhea, rapid and accurate sales data are key. More work is needed in creating actionable algorithms that make use of this data and in understanding the generalizability of our findings to other settings.

A Comparative Analysis of Vibrio cholerae Contamination in Point-of-Drinking and Source Water in a Low-Income Urban Community, Bangladesh

“…The study was conducted in Arichpur, located in Tongi Township of Dhaka, Bangladesh. Arichpur is an urban community with an area of 1.2 km 2 , population density of more than 100,000 residents per km 2 , and approximately 129,000 residents living in 29,000 households (Azman et al, 2015 ). Residents of this area use water from two types of communal pumps: “WASA (Water Supply and Sewerage Authority) pump” installed by the municipal government and connected to households through underground networks of pipes, and/or “submersible pump” installed by individuals or groups of residents and connected to households through over ground networks of pipes.…”

Water usage, hygiene and diarrhea in low-income urban communities—A mixed method prospective longitudinal study

“…1) [5]. Most of the residents of this area were living within a low-income setting and was known to have high incidence of water-borne diseases including cholera [6,7]. This study adopted a mixed method approach to get in-depth understanding on water usage and diarrheal disease of the community people.…”

Trends in the Mechanistic and Dynamic Modeling of Infectious Diseases

“…Similar approaches have been used with Wikipedia updates [68] and social media sites such as Twitter and Facebook. Electronic medication sales data and electronic medical records have also been proposed as novel data sources of disease trends [69]. These approaches can provide estimates much faster than traditional surveillance systems, where it often takes weeks or months for results of cases to be aggregated and analyzed.…”

“…Mechanistic models can then be fit to this data to better understand seasonal or spatial parameters. For example, Yang et al used mechanistic models fit to Google Flu Trend data to estimate epidemiological parameters such as the basic reproductive number and the attack rate for 115 cities in the USA over a 10-year period [69,70].…”