Exploring the relationship between safety culture and reported dispensing errors in a large sample of Swedish community pharmacies

A cross-sectional study was performed, encompassing 546 (62.8%) of the 870 Swedish community pharmacies. All staff in the pharmacies on December 1st, 2007 were included in the study. To assess safety culture domains in the pharmacies, the Safety Attitudes Questionnaire (SAQ) was used. Numbers of dispensed prescription items as well as dispensing errors for each pharmacy across the first half year of 2008 were summarised. Intercorrelations among a number of variables including SAQ survey domains, general properties of the pharmacy, demographic characteristics, and dispensing errors were calculated. A negative binomial regression model was used to further examine the relationship between the variables and dispensing errors.The first analysis demonstrated a number of significant correlations between reported dispensing errors and the variables examined. Negative correlations were found with SAQ domains Teamwork Climate, Safety Climate, Job Satisfaction as well as mean age and response rates. Positive relationships were demonstrated with Stress Recognition (SAQ), number of employees, educational diversity, birth country diversity, education country diversity and number of dispensed prescription items. Variables displaying a significant relationship to errors in this analysis were included in the regression analysis. When controlling for demographic variables, only Stress Recognition, mean age, educational diversity and number of dispensed prescription items and employees, were still associated with dispensing errors.This study replicated previous work linking safety to errors, but went one step further and controlled for a variety of variables. Controlling rendered the relationship between Safety Climate and dispensing insignificant, while the relationship to Stress Recognition remained significant. Variables such as age and education country diversity were found also to correlate with reporting behaviour. Further studies on the demographic variables might generate inte