Getting the Most from Your Surveillance Data: Methods Used By the Massachusetts Sharps Injury Surveillance System

BACKGROUND:  A challenge for the surveillance of occupational injuries and illnesses is that even with comprehensive data about the incidence of these injuries and illnesses it can be difficult to assess the risk from certain events without corresponding denominator information about the frequency of these events. For sharps injury (SI) surveillance, the effectiveness of sharps with engineered sharps injury protections (SESIPs) compared to sharps without those protections (non-SESIPs) cannot be assessed without counts of the devices being used to calculate rates. The Massachusetts Sharps Injury Surveillance System collects SI data from every hospital licensed by the Massachusetts Department of Public Health, but does not collect data about the number of devices used. This study explores methods for assessing how the use of SESIPs has affected SI rates in the absence of this device based denominator. 

METHODS:  We compared SIs per 1,000 FTEs over time for SESIPs and non-SESIPs using negative binomial and joinpoint regression. Next, for specific devices, we compared the rate of SIs over time to the proportion of SIs occurring with SESIPs or non-SESIPs. Finally, we compared the proportion of SESIP and non-SESIP SIs occurring after the device was used (when sharps injury prevention mechanism would be engaged) to those occurring during use (when mechanism is not engaged). 

RESULTS:  Negative binomial regression showed that SESIP rates increased while non-SESIP rates decreased over time. Joinpoint regression showed that SESIP SIs increased from 2002-2005 and have remained steady since 2005, while non-SESIP rates decreased steadily since 2002. For hypodermic needles/syringes and winged steel needles, the rate of SIs declined as the proportion of SESIP SIs increased. The proportion of SIs involving SESIPs has increased at a faster rate for SIs occurring during use compared to after the use of the device. 

CONCLUSIONS:  The increasing rate of SIs from SESIPs is likely due to higher usage of those devices. These findings were used to infer that as SESIPs replaced non-SESIPs, the overall injury rate declined, even as the injury rate for SESIPs themselves stayed level. By evaluating results from a variety of statistical tests, it was possible to use our surveillance data to make inferences about the effectiveness of certain devices even in the absence of device specific denominator information. For other surveillance systems, these findings highlight methods that can be used to explore the risk of occupational illnesses and injuries from different events without ideal denominators.