Summary
Clinician Decision Support Systems (CDSS) provide HCPs with automated advice about best practice patient care, and have been shown to effectively influence behaviour in terms of adherence to guidelines and avoidance of drug errors. The Evidence Based Medicine electronic Decision Support (EBMeDS) system is a CDSS that was developed in conjunction with health care providers and successfully implemented within SCI-Diabetes, the Scottish national electronic health record for diabetes. EBMeDS has been 'live' to users within NHS Tayside and NHS Lothian since December 2013: evaluation found that use of the EBMeDS system has resulted in a dramatic improvement in adherence to national guidelines, with modest improvements in glycaemic control. Future work will aim to develop and implement additional rule-based algorithms based on user feedback and roll out CDSS to all users of SCI-Diabetes across NHS Scotland.
Results
With regards to clinical outcomes, the presence of an EBMeDS message was associated with small improvements in glycaemic control – which are associated with considerable long-term savings due to reduced complications. For example, if the current UK type 1 diabetes population reduced their HbA1c by 4mmol/mol (0.4%) the estimated saving is £39m over five years (£995m after 25 years). QPIs were based on national guidelines e.g. foot screening is recommended annually - QPI was the percentage of patients who have received foot screening in the past 15 months. For each QPI, the number of patients who received screening following a clinic appointment was compared to the control population. The odds (adjusted for demographic factors) of a patient receiving a screening investigation following a clinic appointment were significantly increased. For example, patients in the intervention group who were due foot screening were approximately three times more likely to receive this after their appointment, compared to controls attending clinic elsewhere. Clinical outcomes included: glycaemic control (HbA1c); cardiovascular risk factors (blood pressure and cholesterol); and kidney function (urinary protein and serum creatinine). Overall, the analysis demonstrated small but significantly greater improvement in HbA1c within cases compared to controls. This resulted in greater numbers falling within the target range for HbA1c within the study population. Kidney function deteriorated in cases and controls, however this deterioration was significantly less marked in cases. The use of the EBMeDS system had no adverse effects on patient experience, clinic consultation or working practices. Quantification of system usage supports users' assertion that the system encourages more efficient working practices.
Challenge
The Scottish Diabetes Action plan emphasises the need for a 'person-centredness' approach to support people with diabetes in managing their own condition. Diabetes care in Scotland relies on a series of managed clinical networks supported by a national informatics platform - SCIDiabetes. Regional and national audits are regularly published using this data on a series of quality performance indicators (QPIs), which allow regional and international comparisons to be drawn. Despite some encouraging trends in adherence to QPIs, there is clearly room for improvement.
Objectives
To implement decision support scripts within the SCI-Diabetes system in NHS Tayside and West Lothian, with a view to informing recommendations for a national decision support system. To demonstrate that there are no unintended adverse effects, and to quantify changes in clinical processes and/or outcomes.
Solution
When a patient's electronic health record (EHR) is opened, coded data is sent to the EBMeDS engine consisting of a series of algorithms (or 'scripts'). Results are transmitted back to the EHR for the user to view. Scripts aligned to diabetes care in Scotland were selected. Significant amendments were required to conform to national guidelines and local context. Initial prompt consists of a short decision support message. Navigating to a "long message" provides further details and a hyperlink to the relevant evidence that can be emailed for future reference. The appearance/behaviour of the system was decided in consultation with users and underwent a number of iterations. A 'pop-up' dialogue box that automatically disappears was felt to offer the correct balance between improving access whilst avoiding user fatigue. The SCI-Diabetes architecture allows EBMeDS to be switched on to specific user groups. Awareness raising/training of HCPs was undertaken via departmental meetings. Questionnaires were distributed and focus groups were held prior to implementation that followed a phased approach.
Learnings
For clinical processes, the odds of a patient receiving an appropriate screening investigation following a clinic appointment were significantly raised. Aside from the health benefits afforded by early intervention, there are considerable potential economic savings in areas such as foot screening (where it is possible to extrapolate health savings of £1m per year for NHS Scotland).
Evaluation
Evaluation adopted a mixed methods approach, based upon the NHS Scotland Knowledge into Action framework. The ambition of this strategy is to embed knowledge in care processes in real-time, making the EBMeDS system an ideal exemplar of this approach. Two 'improvement cycles' ran over the course of an 18-month period: phase one at Ninewells Hospital and one primary care practice in Dundee (16 weeks); and phase two in the Tayside area and St John's Hospital, Livingston (16 weeks). A post-implementation questionnaire and focus groups were held towards the end of each improvement cycle. Patient Reported experience measures (PREMs) were collected via questionnaires distributed during each improvement cycle.
