Table 3

Examples of work involved in repurposing data categorized by the problem it illustrates and its source
Example Problem Source
Doctors are asked to enter diagnosis at the time of a patient’s admission. Diagnosis is not always known at this time and of little importance to the doctors, who base their work on active issues. For this reason, it often does not get filled in correctly or at all. Its completion is an augmentation of data needed for direct clinical care. Data entry needs augmenting Data entry not needed for direct clinical care
Junior doctors are asked to enter past medical history twice: into the admission form and into an audit form. They find this frustrating and time consuming and assign it a low priority with the ramification that it does not always get done. Duplicate data entry needed
Nurses are asked to validate measures captured automatically from the monitors. This ensures that values are accurate and not distorted (for example, because the patient moved). Their accuracy is of little value to nurses who can clearly see they are wrong and the correct values adjacent. They do not always correct mistakes, especially if busy. This causes considerable problems at audit time when highest and lowest parameters are searched for. Quality control of data needed
An audit for septic patients is not possible because it is not a diagnosis. The computer can only identify those with bacterial pneumonia as a main diagnosis. In contrast, a nurse can easily tell if a patient is septic from looking at the clinical record. Data not specifically recorded Data not machine-readable
Clinicians prefer to write their notes in free text to more aptly express the issues of focus to their colleagues. This can make diagnosis, problems, and actions difficult to extract from CIS data. Data not in appropriate format
Doctors use a range of terminology for common problems. For example, out of hospital arrest can be termed: cardiac arrest, cardiac standstill, cardiac asystole, and ventricular fibrillation. This makes it very difficult for the computer to search for data related to out of hospital arrest as it cannot assimilate all of the related terms easily as a doctor could do. Terminology not consistent

Supporting data is provided in Additional file 1, CIS, clinical information system.

Morrison et al.

Morrison et al. BMC Medicine 2013 11:103   doi:10.1186/1741-7015-11-103

Open Data