Decision support for diabetes: embedding knowledge in care processes

In this paper we describe the development, implementation and initial evaluation of a Clinical Decision Support System within a national Electronic Health Record for people with diabetes.

1. INTRODUCTION
Long term conditions affect one in five people, and account for 80% of general practice consultations in Scotland [Auditor General for Scotland 2007]. Approximately half of all clinical decisions made by Health Care Practitioners (HCPs) fail to take account of the best available evidence [McGlynn and Asch 2003] and guidelines often do not accommodate co-morbidities and multiple medications [Lugtenberg et al. 2011; Nobili et al. 2011]. There is a recognised need to find innovative ways of integrating knowledge into clinical workflow; to contextualise and personalise care; and to manage the complex care needs and human factors which contribute to unwanted variation in practice.
Clinical decision support systems (CDSS) within an electronic health record (EHR) provide HCP’s with automated, tailored, evidence-based advice. This project aims to implement and evaluate the Evidence Based Medicine electronic Decision Support (EBMeDS) system [Duodecim Medical Publications Ltd. 2014] within SCI-Diabetes, the national EHR for diabetes in Scotland [Cunningham et al. 2011].

2. METHODS
EBMeDS utilises structured patient data from EHRs and provides automated reminders, therapeutic suggestions and diagnosis-specific links to guidelines and literature [Duodecim Medical Publications Ltd. 2014]. EBMeDS scripts were adapted to the Scottish context and integrated within SCI-Diabetes. Implementation is following a phased approach - phase 1 involves NHS Tayside secondary care, phase 2 will include NHS Lothian and primary care .
Ongoing evaluation is based upon the NES knowledge into action framework [NHSScotland 2012] and involves: user and patient questionnaires; HCP focus groups; quantitative analysis of usage data; and case control comparisons of guideline adherence and clinical outcomes. User questionnaires were adapted from a previous evaluation of the EBMeDS system [Heselmans et al. 2012] which utilised the Unified Theory of User Acceptance of Technology (UTAUT) model [Venkatesh et al. 2003].
3. RESULTS
19 EBMeDS scripts were developed for a variety of clinical situations e.g. optimising glycaemic control; uptake of screening services. Alerts and reminders are displayed to users on opening the clinical record of an individual patient record.
Phase 1 commenced December 2013 and involved 24 HCP’s within the diabetes clinic (approximately 500 patients/month). Questionnaire and focus group feedback suggests that users are receptive to using CDSS. However, self-reported system use is minimal. Barriers to adoption include: clinical time; low relevance to the secondary care context; and limited applicability to individual patient circumstances and co-morbidities. There were no reported adverse effects, with high patient satisfaction recorded during the period of evaluation. Scripts have been amended in light of user feedback – thresholds have been altered; additional rules created; and additional user-control has been introduced. Quantitative data analysis of user navigation data and quality performance indicators is ongoing.
4. DISCUSSION
This service improvement project involves the implementation and evaluation of a CDSS. The potential of the system is acknowledged but needs to adapt in response to user feedback. Script development is ongoing with a view to phase 2 implementation in August 2014. The ultimate aim is to develop a national system taking into account patient co-morbidities and clinical context.
5. ACKNOWLEDGMENTS
The authors would like to acknowledge the ongoing support of the following collaborators in the project: Duodecim Medical Publications Ltd; Digital Health Institute (sponsors); NHS Tayside and Lothian.