Article Preview
TopIntroduction
Spreadsheets usage is widespread across all sectors of commercial organizations (Pemberton and Robson 2000, Kruck 2006), most notably in the financial sector (Maditinos et al. 2012, Burdick 2008, Ayalew et al. 2000, Read & Batson 1999, Baxter 2012). Not only are spreadsheets prevalent but they are also commonly of high impact, being used as a key tool in organizations’ operational activities and financial reporting (PWC 2004). Kruck (2006), for example, states that “business professionals use spreadsheets extensively” and that “decisions made using spreadsheets involve billions of dollars”. Their usage in the healthcare sector is less reported, but early work by Croll and Butler (2006) suggests that spreadsheets are prevalent in that domain also.
However, spreadsheet development and usage is error-prone with spreadsheet errors considered “common and non-trivial” (Panko 2000). Reviews of the literature in Kruck (2006) and Panko (1998), show that the numbers of operational and laboratory spreadsheets with errors range between 7% and 82%, with an average of 40% of professional spreadsheets containing errors. KPMG, an international auditing company, (cited in Rajalingham et al. 2000), found over five errors in 95% of the financial models they reviewed. Likewise, Coopers and Lybrand found that 90% of the spreadsheets they audited had errors (Rajalingham et al. 2000) and the “Tuck School of Business at Dartmouth found that 94% of spreadsheets and 5% of all formulae within spreadsheets contain errors” (Arnott 2011). One of the most frequently mentioned reasons for the high error rate in spreadsheets, is the fact that spreadsheets are developed by end-users, having little formal education in computing and lacking knowledge of structured design methods (Ahmad et al. 2003, Powell et al. 2009a).
Information about the actual impact of spreadsheet error is scarce, but severe when reported: Powell et al. (2009a) reported on several instances of erroneous spreadsheets in the financial sector having stark financial impact, with one organisation suffering a $100 million impact. Additionally, the EuSpRIG website reports many anecdotal examples of very poor outcomes due to spreadsheet error (O’Beirne et al. 2013). But overall, there is little academic literature reporting of the risk or impact of these errors, probably due to the sensitivity of the information. Regardless, it can be assumed, that spreadsheets are not used as efficiently and reliably as they could be.
Given the reported prevalence and probable low spreadsheet quality across sectors, the focus of this research has been to investigate error in spreadsheets. The decision to focus on the Healthcare sector arises from the reported prevalence of spreadsheet use in this domain Croll & Butler (2006), and the potentially serious consequences of errors potentially introduced by spreadsheet development and use in this domain: incorrect or poor clinical decisions (resulting in patient harm), loss of finances or resources for appropriate healthcare agendas, or reputational harm. It is therefore important to identify the prevalence of various error types in healthcare spreadsheets so that the problem can be quantified and remedial actions taken to reduce the error rate. Healthcare, in the context of this research, refers to acute hospitals delivering frontline care. Croll & Butler (2006) demonstrated significant errors in three open spreadsheets used for clinical decision making and service planning which could have resulted in significant concerns for patient safety. But apart from this one study, our literature review found little other research into spreadsheet error with respect to these two concerns.
This research moves towards addressing this lack of empirical evidence, focusing on the service planning aspect of healthcare and specifically addressing the following research questions: