Cumulative effects of heat exposure and storage conditions of Oxytocin-in-Uniject in rural Ghana: Implications for scale up

Objective: Postpartum hemorrhage can be reduced substantially in home deliveries attended by community-based workers by using Oxytocin-in-Uniject (OIU) devices affixed with temperature-time indicators. We characterized the distribution of time to discard of these devices when stored under normal field conditions in Ghana. Methods: Two drug storage simulation studies were conducted in rural Ghana in 2011 and 2012. Devices were transported under refrigeration from manufacture (Argentina) to storage at the study site. Twenty-three field workers each stored at home (unrefrigerated) 25 OIU devices and monitored them daily to record: (1) time to transition from usable to unusable, and (2) continuous digital ambient temperature to determine heat exposure over the simulation period. Time to discard was estimated and compared with mean kinetic temperature exposure of the devices during the shipment and storage phases and with characteristics of the storage locations using Weibull regression models. We used the time to discard distributions in a Monte Carlo simulation to estimate wastage rates in a hypothetical program setting. Results: Time for shipment and transfer to long-term refrigerated storage and mean kinetic temperature during the shipment phase was 8.6 days/10.3°C and 13.4 days/12.1°C, for the first and second simulation studies, respectively. Median (range) time to discard when stored under field conditions (unrefrigerated) was 43 (6 to 59) days and 33 (14 to 50) days, respectively. Mean time to discard was 10.0 days shorter in the second simulation, during which mean kinetic temperature exposure was 3.9°C higher. Simulating a monthly distribution system and assuming typical usage, predicted wastage of product was less than 10%. Conclusion: The time to discard of devices was highly sensitive to small changes in temperature exposure. Under field conditions typical in rural Ghana, OIU packages will have a half-life of approximately 30 to 40 days based on the temperature monitor used during the study. Program managers will need to carefully consider variations in both ambient temperature and rate of use to allocate the appropriate supply level that will maximize coverage and minimize stock loss.