Data Driven Solutions for Responsible Medicines use in Secondary Care Study

Abstract

Executive Summary Medicines are a foundation of modern medicine. While they are associated with substantial benefits they are also associated with potentially significant adverse events; many of which are preventable. There is no single contributing factor for medication errors and they can occur across the continuum: from the time of prescribing to administration. The rapid growth and use of electronic prescribing and administration (ePA) systems is suggested as having the greatest potential for improving medication use, significantly contribute to patient safety and enable a new dimension for medicine intelligence. In other words the advent of various ePA systems has the potential to provide new opportunities for exploring medicines use at a population level. The purpose of this case study was to examine and where possible link, three DHB’s ePA systems, other clinical data such as laboratory data and national health data collections, to understand the potential for driving data solutions that support responsible medicine use in secondary care. Waitemata, Counties Manukau and Taranaki DHBs were selected as the three cases. Key informant interviews were undertaken to explore the use of ePAs and the data related benefits experienced, plus any issues that may be impacting on a greater realisation of these benefits. From these interviews, antibiotic use was identified as a focus area for exploring how ePA data could be linked to other data sets to create new insights into medicines use. A data extraction protocol was created. The extraction protocol involved identifying all NHIs with at least one prescribed / administered antibiotic between 1 January 2015 and 30 June 2016. All associated visit information (in patient or ED visit) and medicine information (drug type, dose, frequency, and route) data was to be extracted and submitted to the MoH for encryption. Unfortunately, Taranaki were unable to participate beyond the key informant phase as they acknowledged that their data analysts were inexperienced with the ePA data structures and that it would be a substantive effort for them to become familiar enough to be able to effectively and robustly extract data. Therefore, the exploratory processes and analyses were only undertaken in partnership with the Counties Manukau and Waitemata DHBs. The main DHB systems that data was extracted from were Ascribe and Pyxis and their patient administration system. Neither DHB were able to provide laboratory data at the time and the process with which requesting this information from the regional repository TestSafe would have significantly delayed the study. With regard to national collections, the NMDS, NNPAC and community dispensing (national pharmaceutical warehouse) data sets were also obtained and matched to the DHB data via the MoH encryption service. Early Insights The qualitative information gathering provided the researchers with a clear view of the diversity of data formats and structures, plus the variation in types of data used and for what purpose, within each DHB. The concept maps (pages 23, 27, 32 and 67), illustrate the depth of consideration the key informants put into thinking about the ePA data opportunity. Weighting of level of importance was not done, but this piece of work provides a significant foundation with which a Delphi approach to determining where next is possible. Overall, there were obvious system and corresponding data personalisation (Table 7: p. 47). Current use of data from these systems was described as primarily used for “administrative purposes”. While there were other forays into the data, these were topic led more than a strategic data management and knowledge generation plan. However, as these systems are still relatively new, the paucity of a wider focus on data intelligence is not unusual. It was also evident that the three DHBs had different information specialist capability; as evidenced by the withdrawal of Taranaki form the data part of the study. This is also driven by the difficulty to share codes and queries due to the different bespoke changes. Of particular worry to respondents was a (perceived) lack of national oversight related to the ePA implementations (and local modifications) and that as yet, no implemented ePA had managed to integrate the NZ universal list of medicines (NZULM). As these systems do become more ubiquitous the need for a concomitant focus minimising minor DHB variations (where not clinically relevant) and on developing people with good health informatician skills. What was the data like? This is an exploratory study and as such the findings need to be kept in context of both the environment that the systems are in and the newness of these electronic systems in general. Notwithstanding the differences made by the individual DHBs, the key finding was that much of these could be dealt with at the data management end; thereby minimising changes needed at the DHB end. However, is will be essential that DHBs shift into the digital age when it comes to documentation. There was a distinct paucity of what was changed, what data fields mean, how data was collected and stored etc. A key step will be to begin and maintain a robust documentation system where data dictionaries, coding protocols, and data structures are kept. Without these, when similar extractions take place, then mapping and cleaning processes remain muddy, and highly time consuming. This scoping study intentionally requested each DHB to send the data “as it comes” so that these elements could be explored with regard to interpretations and definitions. Despite the variations, linking the DHB data to the national collections showed a high degree of matched information (Tables 8 and 9: p. 50). This allowed us to illustrate what analyses using this data was possible (see section 3.2.11 Analyses of DHB and National Health Data Collection Data Linkage, p. 56). The increase in data richness that would be possible when laboratory results is added and discharge medicines is well worth ensuring that data sharing is continued. What did we learn? We believe that this scoping exercise presents a first step in the journey to including eMedicine data into the existing health sector data sets. This is the beginning and as such there are significant steps to create a sustainable and systematic foundation. We suggest that the work presented here is a good start and that the learnings identified - data structure, system functionality, and localisation, transferability through to system knowledge such as organisational capacity and skill development – can be used to populated a strategic pathway for PHARMAC to use in partnership with other agencies such as MoH (MedSafe and Medicine Management groups) and HQSC.